[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6MKAH5AKD1F9C8SN2GZXEVF":3,"skill-same:01M6MKAH5AKD1F9C8SN2GZXEVF":74},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":28,"publishedAt":39,"linkOnly":40,"markdownContent":41,"markdownBody":42,"frontMatter":43,"refName":45,"commitSha":46,"skillDirectory":47,"forksCount":48,"repositoryHtmlUrl":49,"repositoryHomepageUrl":50,"ownerHtmlUrl":51,"ownerBlogUrl":52,"ownerDescription":53,"updatedAt":54,"install":55,"resources":67},"01M6MKAH5AKD1F9C8SN2GZXEVF","skill-intent-contract",null,"Use when starting a complex or ambiguous task that risks scope drift","hashgraph-online\u002Fawesome-codex-plugins","hashgraph-online","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F160371962?v=4","01M6CNB6NRJVNNGN14TR01Q6D3","HOL (Hashgraph Online)","01M6JT1HSKKG9V68KQAQGBA7XP",1575,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fnyldn\u002Fclaude-octopus\u002Fskills\u002Fskill-intent-contract\u002FSKILL.md",[18,23],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",{"categoryNo":24,"categoryName":25,"taxonomyType":26,"taxonomyTypeDesc":27},"01JBSEEDCAT000000000000037","律师与法务","PROFESSION","适用职业",[29,30,31,32,33,34,35,36,37,38],"intent","contract","skill intent contract","use","task","drift","scope","true","when","ambiguous","2026-11-10 10:12:30",false,"---\nname: skill-intent-contract\ndescription: \"Use when starting a complex or ambiguous task that risks scope drift\"\ndisable-model-invocation: true\n---\n\n> **Host: Codex CLI** — This skill was designed for Claude Code and adapted for Codex.\n> Cross-reference commands use installed skill names in Codex rather than `\u002Focto:*` slash commands.\n> Use the active Codex shell and subagent tools. Do not claim a provider, model, or host subagent is available until the current session exposes it.\n> For host tool equivalents, see `skills\u002Fblocks\u002Fcodex-host-adapter.md`.\n\n\n# Intent Contract System\n\n## Purpose\n\nThe intent contract creates a **persistent record of user intent** that:\n- Captures what the user is trying to accomplish\n- Defines success criteria upfront\n- Establishes boundaries and constraints\n- Travels through the entire workflow\n- Validates final outputs against original intent\n\nThis closes the loop between intention and delivery.\n\n\n## Intent Contract Structure\n\nThe intent contract is stored in the current resolved plan run directory as `session-intent.md`. Use `scripts\u002Fplan-storage.sh` to create or recover that directory; never write a loose intent file into `.claude\u002F`.\n\n```markdown\n# Intent Contract\n\n**Created**: [ISO timestamp]\n**Workflow**: [discover\u002Fembrace\u002Freview\u002Fetc.]\n**Status**: [active\u002Fvalidating\u002Fcompleted]\n\n## Job Statement\nWhat the user is trying to accomplish (JTBD framework).\n\n[User's goal in plain language]\n\n## Success Criteria\n\n### Good Enough\n- [Minimum viable success criterion 1]\n- [Minimum viable success criterion 2]\n\n### Exceptional\n- [Excellence criterion 1]\n- [Excellence criterion 2]\n\n## Boundaries\nWhat this should NOT be:\n- [Boundary 1: What to avoid]\n- [Boundary 2: What's out of scope]\n\n## Context & Constraints\n\n**Stakeholders**: [Who needs this to work for them]\n**Existing Assets**: [What to build on]\n**Timeline**: [Time constraints if any]\n**Technical Constraints**: [Platform, language, dependencies]\n\n## Clarifying Context\n[Any answers from the 3-question pattern]\n\n## Task Allocation\n**Risk**: [low | intermediate | high]\n**Initiative**: [human | AI | shared] — who starts and proposes\n**Control**: [human | AI | shared] — who oversees execution as it runs\n**Decision rights**: [human | AI] — who has final say on the outcome\n**AI role**: [none | executor | collaborator | challenger]\n**Execution disposition**: [AI-assisted | human-only | pending-user-decision]\n**Escalation decision**: [not-needed | pending | user's recorded resolution]\n**Resolved AUTONOMY_MODE**: [supervised | semi-autonomous | loop-until-approved | autonomous | not-applicable (contract-only sentinel)]\n\n## Validation Checklist\n- [ ] Meets \"good enough\" criteria\n- [ ] Respects all boundaries\n- [ ] Works for all stakeholders\n- [ ] Builds on existing assets appropriately\n- [ ] Allocation still fits what the task turned out to be\n```\n\n\n## Implementation Instructions\n\n### When to Create Intent Contract\n\nCreate an intent contract when:\n- User invokes a major workflow (`\u002Focto:embrace`, `\u002Focto:discover`, `\u002Focto:plan`)\n- User explicitly asks to \"plan\" or \"set goals\" for a task\n- A workflow requires multiple phases and validation\n\n**Do NOT create for:**\n- Quick, single-action commands\n- Simple file reads or searches\n- Conversational questions\n\n### Step 0: Allocate the work before scoping it\n\nRun this before capturing intent. The question is not how to run the task across\nagents but whether it should be delegated at all, and if so, which parts of the\nauthority go where. Framework: Afroogh, Varshney & D'Cruz (2025), *A Task-Driven\nHuman-AI Collaboration* ([arXiv:2505.18422](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.18422)).\n\n**Classify risk.** Complexity is already scored elsewhere — defer to\n`estimate_complexity` and `classify_cynefin` in `scripts\u002Flib\u002Frouting.sh` rather\nthan re-deriving it. Risk is a separate axis that nothing in the codebase\nmeasures, so judge it here on three questions:\n\n- **Irreversibility** — can the effect be undone, and at what cost?\n- **Consequence** — is anything material at stake: safety, money, data, users?\n- **Accountability** — is a specific person expected to answer for the outcome?\n\n| Risk | Reading |\n|------|---------|\n| **Low** | Reversible, no material consequence, no named accountability. |\n| **Intermediate** | Reversible only at real cost, or consequence is unclear. |\n| **High** | Irreversible, materially consequential, or someone must answer for it. |\n\n**Allocate the three dimensions separately.** They are independent, and treating\nthem as one axis is the mistake this step exists to prevent. People readily hand\nAI the *initiative* on unfamiliar work while keeping control and decision rights\n— an allocation a single autonomy slider cannot express.\n\n- **Initiative** — who starts, proposes, drafts.\n- **Control** — who oversees execution while it runs.\n- **Decision rights** — who has final say on the result.\n\nRecord every outcome explicitly:\n\n| Risk \u002F complexity | Initiative | Control | Decision rights | AI role | Execution disposition | Escalation decision | Mode |\n|---|---|---|---|---|---|---|---|\n| Low \u002F low | AI | AI | AI | executor | AI-assisted | not-needed | `autonomous` |\n| Low \u002F high | shared | human | human | collaborator | AI-assisted | not-needed | `loop-until-approved` |\n| High \u002F low | human | human | human | executor | AI-assisted | not-needed | `supervised` |\n| High \u002F high | human | human | human | challenger | AI-assisted | not-needed | `supervised` |\n\nThe High \u002F high allocation is adversarial: the human leads while AI attacks the\nproposed decision as a deliberate counterweight to the human's own bias. In High \u002F\nlow work, AI may execute only the bounded actions the human directly approves.\n\n**The rule that inverts.** For **intermediate-risk** work where uncertainty is\nhighest, the cited evidence says avoid AI entirely — \"neither as a gatekeeper nor\nas a second opinion\". This contradicts the smooth intuition that middling risk\nimplies middling involvement, and it also sits in tension with the same paper's\nbroader claim that complete human autonomy is rarely justified. That tension is\nreal and unresolved; surface it to the user and let them decide rather than\nquietly picking a side.\n\n**Resolve to a setting.** The workflow engine reads one variable,\n`AUTONOMY_MODE`, with four values:\n\n| Allocation | `AUTONOMY_MODE` |\n|---|---|\n| Human holds control and decision rights, approving each phase | `supervised` |\n| AI runs; human is pulled in on failures and quality gates | `semi-autonomous` |\n| AI runs and iterates; human holds final decision rights | `loop-until-approved` |\n| AI holds all three | `autonomous` |\n\nRecord the three dimensions *and* the resolved mode. The mapping is lossy: one\naxis cannot represent three independent allocations, so a contract that stores\nonly the mode loses the reason it was chosen. That record is what a later\nreviewer needs when the allocation turns out to have been wrong.\n\n`not-applicable` is a persisted, contract-only sentinel for human-only work; it\nis not a fifth runtime value and must not be passed to the workflow engine.\n\nFor intermediate risk, record `Execution disposition: pending-user-decision`.\nRecord `Escalation decision: pending`, then stop before execution.\nAsk the user to choose human-only handling or a specific documented AI allocation.\nRecord their answer, rewrite the Task Allocation fields to match it, and change\n`Escalation decision` to the user's resolution before continuing. For a human-only\nresolution, record `AI role: none` and `Resolved AUTONOMY_MODE: not-applicable`;\nhuman-only `not-applicable` must not be passed to the workflow engine. Resolve to the\nsupported human-only behavior and do not execute AI work in that state.\n\nFor a documented AI-assisted resolution, record every Task Allocation field:\n\n- `Initiative`, `Control`, `Decision rights`, and `AI role` from the chosen allocation\n- `Execution disposition: AI-assisted`\n- `Escalation decision: user's recorded resolution`\n- `Resolved AUTONOMY_MODE`: exactly one of `supervised`, `semi-autonomous`,\n  `loop-until-approved`, or `autonomous`, mapped using the table above\n\nValidate the selected runtime mode and only then execute. The runtime must reject\n`not-applicable` and every unknown or unsupported mode before workflow execution rather\nthan defaulting to autonomous behavior.\n\n### Step 1: Capture Intent\n\nAfter asking the 3 clarifying questions in a workflow, prompt the user to define:\n\n```javascript\nAskUserQuestion({\n  questions: [\n    {\n      question: \"What are you ultimately trying to accomplish?\",\n      header: \"Goal\",\n      multiSelect: false,\n      options: [\n        {label: \"Let me describe it\", description: \"I'll write my own goal statement\"},\n        {label: \"Make a decision\", description: \"Choose between options\"},\n        {label: \"Create deliverable\", description: \"Build something specific\"},\n        {label: \"Understand a problem\", description: \"Research and learn\"}\n      ]\n    },\n    {\n      question: \"What defines success for this?\",\n      header: \"Success\",\n      multiSelect: true,\n      options: [\n        {label: \"Clear recommendation\", description: \"Know what to do next\"},\n        {label: \"Working implementation\", description: \"Code that functions\"},\n        {label: \"Team alignment\", description: \"Everyone understands\"},\n        {label: \"Problem solved\", description: \"Issue is resolved\"}\n      ]\n    },\n    {\n      question: \"What should this NOT be or do?\",\n      header: \"Boundaries\",\n      multiSelect: true,\n      options: [\n        {label: \"Over-engineered\", description: \"Keep it simple\"},\n        {label: \"Incomplete\", description: \"Must be production-ready\"},\n        {label: \"Disconnected\", description: \"Must fit our architecture\"},\n        {label: \"Risky\", description: \"Avoid experimental approaches\"}\n      ]\n    }\n  ]\n})\n```\n\nIf user selects \"Let me describe it\", follow up with a text prompt for their custom goal.\n\n### Step 2: Write Intent Contract File\n\nResolve a unique run directory, then use the Write tool to create its `session-intent.md`:\n\n```bash\nPLAN_STORAGE=\"${CLAUDE_PLUGIN_ROOT:-${CODEX_PLUGIN_ROOT:-${HOME}\u002F.claude-octopus\u002Fplugin}}\u002Fscripts\u002Fplan-storage.sh\"\nOCTO_PLAN_DIR=\"$(\"$PLAN_STORAGE\" current \"$PWD\" 2>\u002Fdev\u002Fnull || \"$PLAN_STORAGE\" create \"$PWD\")\"\ncat > \"${OCTO_PLAN_DIR}\u002Fsession-intent.md\" \u003C\u003CEOF\n# Intent Contract\n\n**Created**: $(date -u +\"%Y-%m-%dT%H:%M:%SZ\")\n**Workflow**: ${WORKFLOW_NAME}\n**Status**: active\n\n## Job Statement\n${USER_GOAL}\n\n## Success Criteria\n\n### Good Enough\n${MIN_SUCCESS_CRITERIA}\n\n### Exceptional\n${EXCEPTIONAL_CRITERIA}\n\n## Boundaries\nWhat this should NOT be:\n${BOUNDARIES}\n\n## Context & Constraints\n\n**Stakeholders**: ${STAKEHOLDERS}\n**Timeline**: ${TIMELINE}\n\n## Clarifying Context\n${THREE_QUESTION_ANSWERS}\n\n## Validation Checklist\n- [ ] Meets \"good enough\" criteria\n- [ ] Respects all boundaries\n- [ ] Works for all stakeholders\nEOF\n```\n\n### Step 3: Reference During Execution\n\nThroughout the workflow, recover the current plan directory with `plan-storage.sh current \"$PWD\"` and periodically read its `session-intent.md` to:\n- Stay aligned with user goals\n- Make decisions consistent with boundaries\n- Keep stakeholders in mind\n\nAt key decision points, explicitly say:\n```\nChecking against intent contract: [reference specific criterion]\n```\n\n### Step 4: Validate at End\n\nWhen the workflow completes, read the resolved plan directory's `session-intent.md` and validate:\n\n**Validation Process:**\n\n1. **Read the intent contract**\n2. **Check each success criterion:**\n   - ✓ Met - explain how\n   - ✗ Not met - explain why and what's needed\n   - ~ Partially met - explain gaps\n\n3. **Check boundaries:**\n   - ✓ Respected - confirm\n   - ✗ Violated - explain what happened\n\n4. **Generate validation report:**\n\n```markdown\n# Validation Report\n\n## Success Criteria Check\n\n### Good Enough Criteria\n- [✓] Criterion 1: [How it was met]\n- [✗] Criterion 2: [Why not met, what's needed]\n\n### Exceptional Criteria\n- [~] Criterion 1: [Partial progress explanation]\n\n## Boundary Check\nAll boundaries respected: [Yes\u002FNo]\n- Boundary 1: [✓\u002F✗] [Explanation]\n\n## Gaps & Next Steps\n[If any criteria not met, list concrete next steps]\n\n## Overall Assessment\n[Summary: Does this fulfill the original intent?]\n```\n\n5. **Present to user:**\n   - Show the validation report\n   - Ask if they want to address any gaps\n   - Update intent contract status to \"completed\" or \"validating\"\n\n### Step 5: Update Intent Contract Status\n\nUpdate the `Status` field in the resolved plan directory's `session-intent.md`:\n- `active` → workflow in progress\n- `validating` → checking against criteria\n- `completed` → all criteria met, boundaries respected\n- `incomplete` → some criteria not met, gaps identified\n\n\n## Integration with Workflows\n\n### Embrace Workflow\n\n```\n1. Ask 3 clarifying questions (scope, focus, autonomy)\n2. Create intent contract\n3. DISCOVER phase (reference intent)\n4. DEFINE phase (reference intent)\n5. DEVELOP phase (reference intent)\n6. DELIVER phase (reference intent)\n7. Validate against intent contract\n8. Present validation report\n```\n\n### Discover Workflow\n\n```\n1. Ask 3 clarifying questions (depth, focus, output)\n2. Create intent contract\n3. Execute multi-provider research\n4. Synthesize findings\n5. Validate against intent contract\n6. Present validation report\n```\n\n### Plan Workflow (Future)\n\n```\n1. Capture comprehensive intent\n2. Create intent contract\n3. Route to appropriate workflows\n4. Execute custom sequence\n5. Validate against intent contract\n6. Present validation report\n```\n\n\n## Example Intent Contract\n\n```markdown\n# Intent Contract\n\n**Created**: 2026-01-21T15:30:00Z\n**Workflow**: embrace\n**Status**: active\n\n## Job Statement\nBuild a user authentication system that our team can implement and maintain.\n\n## Success Criteria\n\n### Good Enough\n- Team understands what to build\n- Clear technical approach selected\n- Security considerations documented\n- Implementation plan with steps\n\n### Exceptional\n- Multiple authentication methods evaluated\n- Security audit performed\n- Code examples provided\n- Integration tests included\n\n## Boundaries\nWhat this should NOT be:\n- Over-engineered with unnecessary features\n- Disconnected from our existing Node.js\u002FExpress stack\n- Experimental or unproven technologies\n\n## Context & Constraints\n\n**Stakeholders**: Development team (5 engineers), Product manager\n**Existing Assets**: Express.js API, PostgreSQL database\n**Timeline**: Need to start implementation next sprint\n**Technical Constraints**: Must work with Express.js, PostgreSQL\n\n## Clarifying Context\n\n**Scope**: Medium feature (multiple components)\n**Focus Areas**: Security, Architecture design\n**Autonomy**: Supervised (review after each phase)\n\n## Validation Checklist\n- [ ] Meets \"good enough\" criteria\n- [ ] Respects all boundaries\n- [ ] Works for all stakeholders\n- [ ] Builds on existing assets appropriately\n```\n\n\n## Benefits\n\n**For Users:**\n- Clear expectations set upfront\n- No forgotten requirements\n- Validation against original goals\n- Closed-loop accountability\n\n**For Workflows:**\n- Clear success criteria to optimize for\n- Boundaries to constrain solutions\n- Context for better decisions\n- Validation framework built-in\n\n\n**Ready to use!** Workflows can now create and validate against persistent intent contracts.\n","> **Host: Codex CLI** — This skill was designed for Claude Code and adapted for Codex.\n> Cross-reference commands use installed skill names in Codex rather than `\u002Focto:*` slash commands.\n> Use the active Codex shell and subagent tools. Do not claim a provider, model, or host subagent is available until the current session exposes it.\n> For host tool equivalents, see `skills\u002Fblocks\u002Fcodex-host-adapter.md`.\n\n\n# Intent Contract System\n\n## Purpose\n\nThe intent contract creates a **persistent record of user intent** that:\n- Captures what the user is trying to accomplish\n- Defines success criteria upfront\n- Establishes boundaries and constraints\n- Travels through the entire workflow\n- Validates final outputs against original intent\n\nThis closes the loop between intention and delivery.\n\n\n## Intent Contract Structure\n\nThe intent contract is stored in the current resolved plan run directory as `session-intent.md`. Use `scripts\u002Fplan-storage.sh` to create or recover that directory; never write a loose intent file into `.claude\u002F`.\n\n```markdown\n# Intent Contract\n\n**Created**: [ISO timestamp]\n**Workflow**: [discover\u002Fembrace\u002Freview\u002Fetc.]\n**Status**: [active\u002Fvalidating\u002Fcompleted]\n\n## Job Statement\nWhat the user is trying to accomplish (JTBD framework).\n\n[User's goal in plain language]\n\n## Success Criteria\n\n### Good Enough\n- [Minimum viable success criterion 1]\n- [Minimum viable success criterion 2]\n\n### Exceptional\n- [Excellence criterion 1]\n- [Excellence criterion 2]\n\n## Boundaries\nWhat this should NOT be:\n- [Boundary 1: What to avoid]\n- [Boundary 2: What's out of scope]\n\n## Context & Constraints\n\n**Stakeholders**: [Who needs this to work for them]\n**Existing Assets**: [What to build on]\n**Timeline**: [Time constraints if any]\n**Technical Constraints**: [Platform, language, dependencies]\n\n## Clarifying Context\n[Any answers from the 3-question pattern]\n\n## Task Allocation\n**Risk**: [low | intermediate | high]\n**Initiative**: [human | AI | shared] — who starts and proposes\n**Control**: [human | AI | shared] — who oversees execution as it runs\n**Decision rights**: [human | AI] — who has final say on the outcome\n**AI role**: [none | executor | collaborator | challenger]\n**Execution disposition**: [AI-assisted | human-only | pending-user-decision]\n**Escalation decision**: [not-needed | pending | user's recorded resolution]\n**Resolved AUTONOMY_MODE**: [supervised | semi-autonomous | loop-until-approved | autonomous | not-applicable (contract-only sentinel)]\n\n## Validation Checklist\n- [ ] Meets \"good enough\" criteria\n- [ ] Respects all boundaries\n- [ ] Works for all stakeholders\n- [ ] Builds on existing assets appropriately\n- [ ] Allocation still fits what the task turned out to be\n```\n\n\n## Implementation Instructions\n\n### When to Create Intent Contract\n\nCreate an intent contract when:\n- User invokes a major workflow (`\u002Focto:embrace`, `\u002Focto:discover`, `\u002Focto:plan`)\n- User explicitly asks to \"plan\" or \"set goals\" for a task\n- A workflow requires multiple phases and validation\n\n**Do NOT create for:**\n- Quick, single-action commands\n- Simple file reads or searches\n- Conversational questions\n\n### Step 0: Allocate the work before scoping it\n\nRun this before capturing intent. The question is not how to run the task across\nagents but whether it should be delegated at all, and if so, which parts of the\nauthority go where. Framework: Afroogh, Varshney & D'Cruz (2025), *A Task-Driven\nHuman-AI Collaboration* ([arXiv:2505.18422](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.18422)).\n\n**Classify risk.** Complexity is already scored elsewhere — defer to\n`estimate_complexity` and `classify_cynefin` in `scripts\u002Flib\u002Frouting.sh` rather\nthan re-deriving it. Risk is a separate axis that nothing in the codebase\nmeasures, so judge it here on three questions:\n\n- **Irreversibility** — can the effect be undone, and at what cost?\n- **Consequence** — is anything material at stake: safety, money, data, users?\n- **Accountability** — is a specific person expected to answer for the outcome?\n\n| Risk | Reading |\n|------|---------|\n| **Low** | Reversible, no material consequence, no named accountability. |\n| **Intermediate** | Reversible only at real cost, or consequence is unclear. |\n| **High** | Irreversible, materially consequential, or someone must answer for it. |\n\n**Allocate the three dimensions separately.** They are independent, and treating\nthem as one axis is the mistake this step exists to prevent. People readily hand\nAI the *initiative* on unfamiliar work while keeping control and decision rights\n— an allocation a single autonomy slider cannot express.\n\n- **Initiative** — who starts, proposes, drafts.\n- **Control** — who oversees execution while it runs.\n- **Decision rights** — who has final say on the result.\n\nRecord every outcome explicitly:\n\n| Risk \u002F complexity | Initiative | Control | Decision rights | AI role | Execution disposition | Escalation decision | Mode |\n|---|---|---|---|---|---|---|---|\n| Low \u002F low | AI | AI | AI | executor | AI-assisted | not-needed | `autonomous` |\n| Low \u002F high | shared | human | human | collaborator | AI-assisted | not-needed | `loop-until-approved` |\n| High \u002F low | human | human | human | executor | AI-assisted | not-needed | `supervised` |\n| High \u002F high | human | human | human | challenger | AI-assisted | not-needed | `supervised` |\n\nThe High \u002F high allocation is adversarial: the human leads while AI attacks the\nproposed decision as a deliberate counterweight to the human's own bias. In High \u002F\nlow work, AI may execute only the bounded actions the human directly approves.\n\n**The rule that inverts.** For **intermediate-risk** work where uncertainty is\nhighest, the cited evidence says avoid AI entirely — \"neither as a gatekeeper nor\nas a second opinion\". This contradicts the smooth intuition that middling risk\nimplies middling involvement, and it also sits in tension with the same paper's\nbroader claim that complete human autonomy is rarely justified. That tension is\nreal and unresolved; surface it to the user and let them decide rather than\nquietly picking a side.\n\n**Resolve to a setting.** The workflow engine reads one variable,\n`AUTONOMY_MODE`, with four values:\n\n| Allocation | `AUTONOMY_MODE` |\n|---|---|\n| Human holds control and decision rights, approving each phase | `supervised` |\n| AI runs; human is pulled in on failures and quality gates | `semi-autonomous` |\n| AI runs and iterates; human holds final decision rights | `loop-until-approved` |\n| AI holds all three | `autonomous` |\n\nRecord the three dimensions *and* the resolved mode. The mapping is lossy: one\naxis cannot represent three independent allocations, so a contract that stores\nonly the mode loses the reason it was chosen. That record is what a later\nreviewer needs when the allocation turns out to have been wrong.\n\n`not-applicable` is a persisted, contract-only sentinel for human-only work; it\nis not a fifth runtime value and must not be passed to the workflow engine.\n\nFor intermediate risk, record `Execution disposition: pending-user-decision`.\nRecord `Escalation decision: pending`, then stop before execution.\nAsk the user to choose human-only handling or a specific documented AI allocation.\nRecord their answer, rewrite the Task Allocation fields to match it, and change\n`Escalation decision` to the user's resolution before continuing. For a human-only\nresolution, record `AI role: none` and `Resolved AUTONOMY_MODE: not-applicable`;\nhuman-only `not-applicable` must not be passed to the workflow engine. Resolve to the\nsupported human-only behavior and do not execute AI work in that state.\n\nFor a documented AI-assisted resolution, record every Task Allocation field:\n\n- `Initiative`, `Control`, `Decision rights`, and `AI role` from the chosen allocation\n- `Execution disposition: AI-assisted`\n- `Escalation decision: user's recorded resolution`\n- `Resolved AUTONOMY_MODE`: exactly one of `supervised`, `semi-autonomous`,\n  `loop-until-approved`, or `autonomous`, mapped using the table above\n\nValidate the selected runtime mode and only then execute. The runtime must reject\n`not-applicable` and every unknown or unsupported mode before workflow execution rather\nthan defaulting to autonomous behavior.\n\n### Step 1: Capture Intent\n\nAfter asking the 3 clarifying questions in a workflow, prompt the user to define:\n\n```javascript\nAskUserQuestion({\n  questions: [\n    {\n      question: \"What are you ultimately trying to accomplish?\",\n      header: \"Goal\",\n      multiSelect: false,\n      options: [\n        {label: \"Let me describe it\", description: \"I'll write my own goal statement\"},\n        {label: \"Make a decision\", description: \"Choose between options\"},\n        {label: \"Create deliverable\", description: \"Build something specific\"},\n        {label: \"Understand a problem\", description: \"Research and learn\"}\n      ]\n    },\n    {\n      question: \"What defines success for this?\",\n      header: \"Success\",\n      multiSelect: true,\n      options: [\n        {label: \"Clear recommendation\", description: \"Know what to do next\"},\n        {label: \"Working implementation\", description: \"Code that functions\"},\n        {label: \"Team alignment\", description: \"Everyone understands\"},\n        {label: \"Problem solved\", description: \"Issue is resolved\"}\n      ]\n    },\n    {\n      question: \"What should this NOT be or do?\",\n      header: \"Boundaries\",\n      multiSelect: true,\n      options: [\n        {label: \"Over-engineered\", description: \"Keep it simple\"},\n        {label: \"Incomplete\", description: \"Must be production-ready\"},\n        {label: \"Disconnected\", description: \"Must fit our architecture\"},\n        {label: \"Risky\", description: \"Avoid experimental approaches\"}\n      ]\n    }\n  ]\n})\n```\n\nIf user selects \"Let me describe it\", follow up with a text prompt for their custom goal.\n\n### Step 2: Write Intent Contract File\n\nResolve a unique run directory, then use the Write tool to create its `session-intent.md`:\n\n```bash\nPLAN_STORAGE=\"${CLAUDE_PLUGIN_ROOT:-${CODEX_PLUGIN_ROOT:-${HOME}\u002F.claude-octopus\u002Fplugin}}\u002Fscripts\u002Fplan-storage.sh\"\nOCTO_PLAN_DIR=\"$(\"$PLAN_STORAGE\" current \"$PWD\" 2>\u002Fdev\u002Fnull || \"$PLAN_STORAGE\" create \"$PWD\")\"\ncat > \"${OCTO_PLAN_DIR}\u002Fsession-intent.md\" \u003C\u003CEOF\n# Intent Contract\n\n**Created**: $(date -u +\"%Y-%m-%dT%H:%M:%SZ\")\n**Workflow**: ${WORKFLOW_NAME}\n**Status**: active\n\n## Job Statement\n${USER_GOAL}\n\n## Success Criteria\n\n### Good Enough\n${MIN_SUCCESS_CRITERIA}\n\n### Exceptional\n${EXCEPTIONAL_CRITERIA}\n\n## Boundaries\nWhat this should NOT be:\n${BOUNDARIES}\n\n## Context & Constraints\n\n**Stakeholders**: ${STAKEHOLDERS}\n**Timeline**: ${TIMELINE}\n\n## Clarifying Context\n${THREE_QUESTION_ANSWERS}\n\n## Validation Checklist\n- [ ] Meets \"good enough\" criteria\n- [ ] Respects all boundaries\n- [ ] Works for all stakeholders\nEOF\n```\n\n### Step 3: Reference During Execution\n\nThroughout the workflow, recover the current plan directory with `plan-storage.sh current \"$PWD\"` and periodically read its `session-intent.md` to:\n- Stay aligned with user goals\n- Make decisions consistent with boundaries\n- Keep stakeholders in mind\n\nAt key decision points, explicitly say:\n```\nChecking against intent contract: [reference specific criterion]\n```\n\n### Step 4: Validate at End\n\nWhen the workflow completes, read the resolved plan directory's `session-intent.md` and validate:\n\n**Validation Process:**\n\n1. **Read the intent contract**\n2. **Check each success criterion:**\n   - ✓ Met - explain how\n   - ✗ Not met - explain why and what's needed\n   - ~ Partially met - explain gaps\n\n3. **Check boundaries:**\n   - ✓ Respected - confirm\n   - ✗ Violated - explain what happened\n\n4. **Generate validation report:**\n\n```markdown\n# Validation Report\n\n## Success Criteria Check\n\n### Good Enough Criteria\n- [✓] Criterion 1: [How it was met]\n- [✗] Criterion 2: [Why not met, what's needed]\n\n### Exceptional Criteria\n- [~] Criterion 1: [Partial progress explanation]\n\n## Boundary Check\nAll boundaries respected: [Yes\u002FNo]\n- Boundary 1: [✓\u002F✗] [Explanation]\n\n## Gaps & Next Steps\n[If any criteria not met, list concrete next steps]\n\n## Overall Assessment\n[Summary: Does this fulfill the original intent?]\n```\n\n5. **Present to user:**\n   - Show the validation report\n   - Ask if they want to address any gaps\n   - Update intent contract status to \"completed\" or \"validating\"\n\n### Step 5: Update Intent Contract Status\n\nUpdate the `Status` field in the resolved plan directory's `session-intent.md`:\n- `active` → workflow in progress\n- `validating` → checking against criteria\n- `completed` → all criteria met, boundaries respected\n- `incomplete` → some criteria not met, gaps identified\n\n\n## Integration with Workflows\n\n### Embrace Workflow\n\n```\n1. Ask 3 clarifying questions (scope, focus, autonomy)\n2. Create intent contract\n3. DISCOVER phase (reference intent)\n4. DEFINE phase (reference intent)\n5. DEVELOP phase (reference intent)\n6. DELIVER phase (reference intent)\n7. Validate against intent contract\n8. Present validation report\n```\n\n### Discover Workflow\n\n```\n1. Ask 3 clarifying questions (depth, focus, output)\n2. Create intent contract\n3. Execute multi-provider research\n4. Synthesize findings\n5. Validate against intent contract\n6. Present validation report\n```\n\n### Plan Workflow (Future)\n\n```\n1. Capture comprehensive intent\n2. Create intent contract\n3. Route to appropriate workflows\n4. Execute custom sequence\n5. Validate against intent contract\n6. Present validation report\n```\n\n\n## Example Intent Contract\n\n```markdown\n# Intent Contract\n\n**Created**: 2026-01-21T15:30:00Z\n**Workflow**: embrace\n**Status**: active\n\n## Job Statement\nBuild a user authentication system that our team can implement and maintain.\n\n## Success Criteria\n\n### Good Enough\n- Team understands what to build\n- Clear technical approach selected\n- Security considerations documented\n- Implementation plan with steps\n\n### Exceptional\n- Multiple authentication methods evaluated\n- Security audit performed\n- Code examples provided\n- Integration tests included\n\n## Boundaries\nWhat this should NOT be:\n- Over-engineered with unnecessary features\n- Disconnected from our existing Node.js\u002FExpress stack\n- Experimental or unproven technologies\n\n## Context & Constraints\n\n**Stakeholders**: Development team (5 engineers), Product manager\n**Existing Assets**: Express.js API, PostgreSQL database\n**Timeline**: Need to start implementation next sprint\n**Technical Constraints**: Must work with Express.js, PostgreSQL\n\n## Clarifying Context\n\n**Scope**: Medium feature (multiple components)\n**Focus Areas**: Security, Architecture design\n**Autonomy**: Supervised (review after each phase)\n\n## Validation Checklist\n- [ ] Meets \"good enough\" criteria\n- [ ] Respects all boundaries\n- [ ] Works for all stakeholders\n- [ ] Builds on existing assets appropriately\n```\n\n\n## Benefits\n\n**For Users:**\n- Clear expectations set upfront\n- No forgotten requirements\n- Validation against original goals\n- Closed-loop accountability\n\n**For Workflows:**\n- Clear success criteria to optimize for\n- Boundaries to constrain solutions\n- Context for better decisions\n- Validation framework built-in\n\n\n**Ready to use!** Workflows can now create and validate against persistent intent contracts.\n",{"name":5,"description":7,"disable-model-invocation":44},true,"main","f620abaff09061f8c8b1f1559dca3fdd031a32e5","plugins\u002Fnyldn\u002Fclaude-octopus\u002Fskills\u002Fskill-intent-contract",491,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins","https:\u002F\u002Fhol.org\u002Fplugins\u002Fbest-codex-plugins","https:\u002F\u002Fgithub.com\u002Fhashgraph-online","https:\u002F\u002Fhol.org","Open-source standards and developer tools for interoperable AI agents: discovery, identity, communication, payments, privacy and security.","2026-11-14 14:06:13",{"source":8,"skillName":5,"commands":56,"prompt":66,"zipAvailable":44},[57,60,63],{"tool":58,"command":59},"npx","npx skills add hashgraph-online\u002Fawesome-codex-plugins --skill skill-intent-contract",{"tool":61,"command":62},"bunx","bunx skills add hashgraph-online\u002Fawesome-codex-plugins --skill skill-intent-contract",{"tool":64,"command":65},"pnpm","pnpm dlx skills add hashgraph-online\u002Fawesome-codex-plugins --skill skill-intent-contract","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fnyldn\u002Fclaude-octopus\u002Fskills\u002Fskill-intent-contract\u002FSKILL.md",[68],{"path":69,"resourceType":70,"resourceTypeDesc":71,"sizeBytes":72,"sourceHtmlUrl":73},"agents\u002Fopenai.yaml","REFERENCE","参考文档",171,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fnyldn\u002Fclaude-octopus\u002Fskills\u002Fskill-intent-contract\u002Fagents\u002Fopenai.yaml",[75,85,93,101,109,117],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":79,"licenseFileUrl":80,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":84},"01M7DV332YQMGD8WTRJ9DZQXTB","worktable-review-with-annotations","Create, inspect, reply to, update, or resolve situated Worktable annotations. Use when feedback belongs to a Doc section, block, text range, or HTML Doc, or when the user asks to address existing review feedback and coordinate a requested artifact edit.","MIT","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-review-with-annotations\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-review-with-annotations\u002FSKILL.md",[],[],"2026-11-15 14:19:17",{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":79,"licenseFileUrl":89,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":84},"01M7DV2YGHCZVBEB1AHYB6CGHE","worktable-find-and-synthesize-context","Find, assess, and synthesize context from Worktable Docs and Records. Use when the user asks what a workspace contains, wants workspace-backed research, needs current facts reconciled across artifacts, or requests a cited synthesis without changing workspace state.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-find-and-synthesize-context\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-find-and-synthesize-context\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":79,"licenseFileUrl":97,"sourceHtmlUrl":98,"categories":99,"keywords":100,"publishedAt":84},"01M7DV2TQN1H11ME2SAF9MXN7V","worktable-create-or-update-docs","Create, edit, rename, or maintain narrative Worktable Docs. Use for plans, decisions, briefs, notes, research syntheses, and other durable prose when the user requests a new artifact or a safe update to an existing one.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-create-or-update-docs\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-create-or-update-docs\u002FSKILL.md",[],[],{"skillNo":102,"name":103,"nameZh":6,"description":104,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":79,"licenseFileUrl":105,"sourceHtmlUrl":106,"categories":107,"keywords":108,"publishedAt":84},"01M7DV2NE452NEA8TXATKSNAJH","worktable-create-or-manage-records","Organize repeated structured information as durable Worktable Records. Use for research datasets, inventories, directories, pipelines, trackers, and other collections whose items need identity, fields, filtering, links, validation, or individual updates.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-create-or-manage-records\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-create-or-manage-records\u002FSKILL.md",[],[],{"skillNo":110,"name":111,"nameZh":6,"description":112,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":79,"licenseFileUrl":113,"sourceHtmlUrl":114,"categories":115,"keywords":116,"publishedAt":84},"01M7DV2HENRBR2KSCVM064WH0Y","worktable-create-or-manage-interactive-html","Create or improve durable visual and interactive artifacts in Worktable. Use for dashboards, trackers, calculators, mini apps, slide presentations, and other requests that benefit from a polished HTML Doc.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-create-or-manage-interactive-html\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-create-or-manage-interactive-html\u002FSKILL.md",[],[],{"skillNo":118,"name":119,"nameZh":6,"description":120,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":79,"licenseFileUrl":121,"sourceHtmlUrl":122,"categories":123,"keywords":124,"publishedAt":84},"01M7DV2CADR5R7RQFFYPNAGR9Y","worktable-collaborate-threads","Read, answer, and hand off messages through durable Worktable threads with users or other authenticated participants. Use when the user asks to check addressed messages, reply to a participant, continue a thread, request another agent's help, wait briefly for a response, or watch for comments and messages sent to the agent.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-collaborate-threads\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fworktable\u002Fworktable-plugin\u002Fskills\u002Fworktable-collaborate-threads\u002FSKILL.md",[],[]]