[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77WD3CY7XFNE7GRW214ZEDP":3,"skill-same:01M77WD3CY7XFNE7GRW214ZEDP":47},{"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":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M77WD3CY7XFNE7GRW214ZEDP","agent-access",null,"Give your own agent (OpenClaw or Hermes Agent today, more to come) its own identity, scoped and paid access to your data, and a way to cut it off. Use when: (1) the user says \"give my agent access\", \"let OpenClaw read my data\", \"connect Hermes to Vana\", \"set up an agent in a sandbox with Vana\", (2) an agent the user runs should see some of their data but not act as them. This skill is the owner's side. The agent itself uses the builder skill. For collecting data, use connect-data.","vana-com\u002Fvana-cli","vana-com","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F123112561?v=4","01M77WCAG7S42PDY7VTEVWXQD1","Vana","01M77WCAGV436KWGZWHNHZXGVH",2,"MIT","https:\u002F\u002Fgithub.com\u002Fvana-com\u002Fvana-cli\u002Fblob\u002Fmain\u002Fskills\u002Fagent-access\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: agent-access\ndescription: >\n  Give your own agent (OpenClaw or Hermes Agent today, more to come) its\n  own identity, scoped and paid access to your data, and a way to cut it\n  off. Use when: (1) the user says \"give my agent access\", \"let OpenClaw\n  read my data\", \"connect Hermes to Vana\", \"set up an agent in a sandbox\n  with Vana\", (2) an agent the user runs should see some of their data but\n  not act as them. This skill is the owner's side. The agent itself uses the\n  builder skill. For collecting data, use connect-data.\n---\n\n# Agent Access\n\nReading your own data is free, because you are the owner. An agent you\nrun should not act as you. This skill sets it up as **its own app**:\n\n- it sees only the scopes the owner approves\n- every read it makes is logged against its grant and paid from its own\n  escrow\n- the owner can revoke it, and its next read is refused\n\nTwo roles, often on two machines:\n\n| Role  | Who                         | Holds                                    |\n| ----- | --------------------------- | ---------------------------------------- |\n| Owner | the person whose data it is | their Vana login and Personal Server     |\n| Agent | the agent the user picks    | its own app key and escrow, nothing else |\n\nNever give the agent the owner's login. Do not copy `~\u002F.vana` from the\nowner's machine to the agent's.\n\n## First, ask which agent\n\nBefore running anything, ask the user which agent they want to give\naccess to, and wait for the answer. If they already named OpenClaw or\nHermes Agent, confirm it instead of asking again.\n\n1. **OpenClaw** ([openclaw.ai](https:\u002F\u002Fopenclaw.ai))\n2. **Hermes Agent** by Nous Research\n   ([hermes-agent.nousresearch.com](https:\u002F\u002Fhermes-agent.nousresearch.com))\n\nTell them that more agents will be supported. Vana is not tied to one\nagent: every agent runs the same `vana` commands and follows the same\nbuilder skill, and only installing the agent and loading the skill differ.\n\nIf they name another agent that can run shell commands, the Vana steps\nbelow still apply. Say that this agent is not a documented path yet, then\nin step 2 install the Vana CLI and the builder skill as usual and make\nthe skill visible to the agent: point its skills directory at\n`~\u002F.agents\u002Fskills`, or copy `~\u002F.agents\u002Fskills\u002Fvana-builder` into the\ndirectory its docs name. If it can't load `SKILL.md` skills at all, give\nit the contents of `vana-builder\u002FSKILL.md` as instructions.\n\nIn the same message, ask:\n\n- which data the agent should see (for example GitHub or Spotify)\n- whether it runs on this machine or in a sandbox (step 3)\n- straight to mainnet (the default, real money) or testnet first (`--network moksha`)\n\n## Stop for the human\n\nThese steps need the person. Say what they need to do, then wait:\n\n1. logging in (`vana login` opens a browser)\n2. connecting a source (`vana connect` can open a browser to sign in to\n   it)\n3. the first `vana server start`, which can ask them to confirm the\n   install and to register the server in a browser\n4. setting up the agent with a model provider key\n5. funding the agent's escrow (real money on mainnet)\n6. approving the agent's request (an approval URL)\n\n## 1. Owner: data and a server the agent can reach\n\nOn the owner's machine:\n\n```bash\nnpm install -g vana-cli             # or: npx vana-cli \u003Ccommand>\nvana login\nvana connect github                 # or any source from `vana sources`\nvana server start\nvana server status --json\n```\n\n`server start` registers a public URL and keeps the Personal Server\nrunning in the background. The agent reads through that URL, so it\nanswers only while the owner's machine is on. If the output says **local\nonly**, the tunnel or the registration failed and the agent cannot reach\nthe server; fix that before going on.\n\nIf `server start` or `server status` says the running server was started\nby an earlier `vana` and must be restarted, do it:\n\n```bash\nvana server stop\nvana server start\n```\n\n## 2. Install the agent and load the builder skill\n\nFollow the section for the agent the user picked.\n\n### OpenClaw\n\n```bash\nnpm install -g openclaw@latest\nopenclaw onboard --install-daemon     # asks for a model provider key\nnpm install -g vana-cli\nvana skills install builder\n```\n\nOpenClaw loads skills from `~\u002F.agents\u002Fskills`, where `vana skills install`\nputs them, unless `OPENCLAW_STATE_DIR` moves its state elsewhere.\n\n### Hermes Agent\n\n```bash\ncurl -fsSL https:\u002F\u002Fhermes-agent.nousresearch.com\u002Finstall.sh | bash\nhermes setup                          # asks for a model provider\nnpm install -g vana-cli\nvana skills install builder\n```\n\nHermes Agent loads skills from `~\u002F.hermes\u002Fskills`, not\n`~\u002F.agents\u002Fskills`. Add the shared directory to `~\u002F.hermes\u002Fconfig.yaml`,\nmerging with any `skills:` section already there:\n\n```yaml\nskills:\n  external_dirs:\n    - ~\u002F.agents\u002Fskills\n```\n\nStart a new Hermes session, then check that `builder` appears in\n`hermes skills list`.\n\n## 3. Choose where the agent runs\n\n- **Same machine**: quickest to try. The agent runs as the owner's user, so\n  it could still reach the owner's login and run owner commands such as\n  `vana data show`. The grant limits what it asks for, not what it can\n  reach.\n- **Sandbox**: the grant becomes the boundary, because the agent holds\n  only its app key. Use this for anything real. If the sandbox can be\n  recreated, it loses the key `register` made in the last one, and the\n  escrow funded to it. Give the agent a fixed key instead: generate it\n  with `echo \"0x$(openssl rand -hex 32)\"`, keep it outside the sandbox,\n  and pass it in as `VANA_APP_KEY`. Never mount the owner's `~\u002F.vana`\n  into the sandbox.\n  - OpenClaw: repeat step 2 inside a container or cloud VM\n    ([docs.openclaw.ai\u002Finstall](https:\u002F\u002Fdocs.openclaw.ai\u002Finstall)).\n  - Hermes Agent: keep it on the owner's machine and set\n    `terminal.backend` in `~\u002F.hermes\u002Fconfig.yaml` to `docker`, `modal`,\n    `daytona` or another sandbox backend, so its shell commands run\n    there. The `vana` commands then run in the sandbox, so install the\n    CLI inside it (the standalone installer needs no Node:\n    `curl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fvana-com\u002Fvana-cli\u002Fmain\u002Finstall\u002Finstall.sh | sh`),\n    and again whenever the sandbox is recreated. Save the fixed key as\n    `VANA_APP_KEY` in `~\u002F.hermes\u002F.env` and forward it:\n\n    ```yaml\n    terminal:\n      backend: docker\n      env_passthrough:\n        - VANA_APP_KEY\n      docker_forward_env: # Docker backend only\n        - VANA_APP_KEY\n    ```\n\nStep 4 runs wherever the agent's shell commands run.\n\n## 4. Agent: identity\n\n```bash\nvana app register --app-url \u003Curl that identifies the agent>\nvana app whoami --json\n```\n\nUse the agent's homepage as the URL, such as\n`https:\u002F\u002Fgithub.com\u002Fopenclaw\u002Fopenclaw` or\n`https:\u002F\u002Fgithub.com\u002FNousResearch\u002Fhermes-agent`.\n\n`register` uses `VANA_APP_KEY` when it is set; otherwise it creates an\napp key separate from the owner's account, in the OS keychain where there\nis one, or in `~\u002F.vana\u002Fapp-key.json`. It is safe to run again. To run\ntwo agents on one machine, give each its own key through `VANA_APP_KEY`.\n`whoami` prints the **app address**; the owner needs it to fund escrow.\n\n## 5. Fund the agent's escrow (human step)\n\nThe simplest path for the person: run\n\n```bash\nvana app claim --json\n```\n\nand send them `data.claimUrl`. They open it while signed in to Vana\nAccount, click **Add and fund**, and land on the agent's Fund page, where\nthey deposit from a browser wallet: USDC.e on mainnet, faucet VANA on\nmoksha. The link is valid for about 50 minutes, and the first Account to\nopen it takes the app, so send it only to the owner. Run `claim` again for\na fresh link. Without the claim, the agent is in nobody's Apps list and\nAccount has no Fund page for it.\n\nFrom the agent's own terminal instead, the agent's key pays the gas, so\nits wallet needs VANA as well as the fee asset:\n\n```bash\nvana app escrow fund --amount 1 --json          # moksha: native VANA\nvana app escrow fund --amount 5 --asset \u003CUSDC.e token address> --network mainnet --yes --json\n```\n\nThe asset to fund is the one reads are priced in; a read without `--pay`\nnames it as `asset` in its exit-4 payload.\n\nThen check:\n\n```bash\nvana app escrow balance --json\n```\n\nFund small amounts. The agent can spend nothing beyond what escrow holds,\nand `--max-fee` caps each read.\n\n## 6. Grant: the agent asks, the owner approves\n\nAsk for the data the user chose. `request` does not check scope names: a\nwrong one still returns an approval URL, and fails only when the owner\nopens it. Get the exact names on the owner's machine, after `vana\nconnect` has collected the source:\n\n```bash\nvana data show github --json    # data.requestedScopes lists the scopes\n```\n\nThen the agent runs:\n\n```bash\nvana app request --scopes github.repositories --no-input --json\n```\n\nThat exits **7** with `approvalUrl`. Hand the URL to the owner. They sign\nin, review the scopes and approve only what the task needs. The agent then\npicks up the grant:\n\n```bash\nvana app requests show \u003Crequest-id> --json\n```\n\n`\u003Crequest-id>` is `data.requestId` from the request. `requests show`\nexits 0 either way: until the owner approves, `data.status` is `pending`\nand `data.grantId` is null. Once approved, `data.grantId` is the grant to\nread with. If the owner later revokes it, or a newer approval for the same\napp replaces its scopes, `data.live` turns false, `data.grant.state` says\n`revoked` or `replaced`, and `next` is a fresh `vana app request`.\n\nKeep requests to sources the Vana web app lists. A request that includes\na source it doesn't list cannot be approved on the web, and data from\nlocal-only sources (such as Claude Code history) can't be granted this way.\n\n## 7. The agent reads and pays\n\n```bash\nvana app read github.repositories --grant \u003Cgrant-id> --pay --max-fee 0.05 --json\n```\n\nWithout `--pay` it stops at exit **4** with the price. Every read is\ncharged, including a re-read of data that hasn't changed, so an agent that\nreads on a schedule spends on every run. The owner's own reads stay free.\n\n## 8. Review and revoke\n\nIn the Vana app, **Settings → Access history** lists each read the agent\nmade under its grant. **Revoke** ends its access: the Personal Server\nrefuses the next read and nothing is charged. To give access again, the\nagent asks again.\n\n## Networks\n\nEverything defaults to **mainnet**, where fees are real USDC.e.\nGrants, balances and server registrations do not carry across networks.\nTo try the whole flow on the testnet first, pass `--network moksha` on every\ncommand (every `vana` command accepts it), `vana server start` included: an\nagent reading on moksha cannot find a server registered only on mainnet.\n\n## Failure notes\n\n- Exit 5 on a read: the owner's Personal Server did not answer. Check\n  `vana server status` on the owner's machine. A read refused after a\n  revoke can also report exit 5 today; if the owner just revoked, treat it\n  as final and ask again rather than retrying.\n- Exit 4 with `--pay` set: `--max-fee` is below the price, or escrow is\n  empty. The limit is in the fee's own asset (USDC.e on mainnet).\n- Exit 7 from `request`: nobody has approved yet. Do not re-request in a\n  loop; poll `requests show` at a human pace.\n- The agent never runs `vana` commands: it did not load the builder\n  skill. Recheck step 2 for the agent the user picked.\n- `delivery: \"enclave\"` with `paid: false`: the owner serves this data\n  from a TEE sandbox, which the gateway currently admits at zero price.\n","# Agent Access\n\nReading your own data is free, because you are the owner. An agent you\nrun should not act as you. This skill sets it up as **its own app**:\n\n- it sees only the scopes the owner approves\n- every read it makes is logged against its grant and paid from its own\n  escrow\n- the owner can revoke it, and its next read is refused\n\nTwo roles, often on two machines:\n\n| Role  | Who                         | Holds                                    |\n| ----- | --------------------------- | ---------------------------------------- |\n| Owner | the person whose data it is | their Vana login and Personal Server     |\n| Agent | the agent the user picks    | its own app key and escrow, nothing else |\n\nNever give the agent the owner's login. Do not copy `~\u002F.vana` from the\nowner's machine to the agent's.\n\n## First, ask which agent\n\nBefore running anything, ask the user which agent they want to give\naccess to, and wait for the answer. If they already named OpenClaw or\nHermes Agent, confirm it instead of asking again.\n\n1. **OpenClaw** ([openclaw.ai](https:\u002F\u002Fopenclaw.ai))\n2. **Hermes Agent** by Nous Research\n   ([hermes-agent.nousresearch.com](https:\u002F\u002Fhermes-agent.nousresearch.com))\n\nTell them that more agents will be supported. Vana is not tied to one\nagent: every agent runs the same `vana` commands and follows the same\nbuilder skill, and only installing the agent and loading the skill differ.\n\nIf they name another agent that can run shell commands, the Vana steps\nbelow still apply. Say that this agent is not a documented path yet, then\nin step 2 install the Vana CLI and the builder skill as usual and make\nthe skill visible to the agent: point its skills directory at\n`~\u002F.agents\u002Fskills`, or copy `~\u002F.agents\u002Fskills\u002Fvana-builder` into the\ndirectory its docs name. If it can't load `SKILL.md` skills at all, give\nit the contents of `vana-builder\u002FSKILL.md` as instructions.\n\nIn the same message, ask:\n\n- which data the agent should see (for example GitHub or Spotify)\n- whether it runs on this machine or in a sandbox (step 3)\n- straight to mainnet (the default, real money) or testnet first (`--network moksha`)\n\n## Stop for the human\n\nThese steps need the person. Say what they need to do, then wait:\n\n1. logging in (`vana login` opens a browser)\n2. connecting a source (`vana connect` can open a browser to sign in to\n   it)\n3. the first `vana server start`, which can ask them to confirm the\n   install and to register the server in a browser\n4. setting up the agent with a model provider key\n5. funding the agent's escrow (real money on mainnet)\n6. approving the agent's request (an approval URL)\n\n## 1. Owner: data and a server the agent can reach\n\nOn the owner's machine:\n\n```bash\nnpm install -g vana-cli             # or: npx vana-cli \u003Ccommand>\nvana login\nvana connect github                 # or any source from `vana sources`\nvana server start\nvana server status --json\n```\n\n`server start` registers a public URL and keeps the Personal Server\nrunning in the background. The agent reads through that URL, so it\nanswers only while the owner's machine is on. If the output says **local\nonly**, the tunnel or the registration failed and the agent cannot reach\nthe server; fix that before going on.\n\nIf `server start` or `server status` says the running server was started\nby an earlier `vana` and must be restarted, do it:\n\n```bash\nvana server stop\nvana server start\n```\n\n## 2. Install the agent and load the builder skill\n\nFollow the section for the agent the user picked.\n\n### OpenClaw\n\n```bash\nnpm install -g openclaw@latest\nopenclaw onboard --install-daemon     # asks for a model provider key\nnpm install -g vana-cli\nvana skills install builder\n```\n\nOpenClaw loads skills from `~\u002F.agents\u002Fskills`, where `vana skills install`\nputs them, unless `OPENCLAW_STATE_DIR` moves its state elsewhere.\n\n### Hermes Agent\n\n```bash\ncurl -fsSL https:\u002F\u002Fhermes-agent.nousresearch.com\u002Finstall.sh | bash\nhermes setup                          # asks for a model provider\nnpm install -g vana-cli\nvana skills install builder\n```\n\nHermes Agent loads skills from `~\u002F.hermes\u002Fskills`, not\n`~\u002F.agents\u002Fskills`. Add the shared directory to `~\u002F.hermes\u002Fconfig.yaml`,\nmerging with any `skills:` section already there:\n\n```yaml\nskills:\n  external_dirs:\n    - ~\u002F.agents\u002Fskills\n```\n\nStart a new Hermes session, then check that `builder` appears in\n`hermes skills list`.\n\n## 3. Choose where the agent runs\n\n- **Same machine**: quickest to try. The agent runs as the owner's user, so\n  it could still reach the owner's login and run owner commands such as\n  `vana data show`. The grant limits what it asks for, not what it can\n  reach.\n- **Sandbox**: the grant becomes the boundary, because the agent holds\n  only its app key. Use this for anything real. If the sandbox can be\n  recreated, it loses the key `register` made in the last one, and the\n  escrow funded to it. Give the agent a fixed key instead: generate it\n  with `echo \"0x$(openssl rand -hex 32)\"`, keep it outside the sandbox,\n  and pass it in as `VANA_APP_KEY`. Never mount the owner's `~\u002F.vana`\n  into the sandbox.\n  - OpenClaw: repeat step 2 inside a container or cloud VM\n    ([docs.openclaw.ai\u002Finstall](https:\u002F\u002Fdocs.openclaw.ai\u002Finstall)).\n  - Hermes Agent: keep it on the owner's machine and set\n    `terminal.backend` in `~\u002F.hermes\u002Fconfig.yaml` to `docker`, `modal`,\n    `daytona` or another sandbox backend, so its shell commands run\n    there. The `vana` commands then run in the sandbox, so install the\n    CLI inside it (the standalone installer needs no Node:\n    `curl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fvana-com\u002Fvana-cli\u002Fmain\u002Finstall\u002Finstall.sh | sh`),\n    and again whenever the sandbox is recreated. Save the fixed key as\n    `VANA_APP_KEY` in `~\u002F.hermes\u002F.env` and forward it:\n\n    ```yaml\n    terminal:\n      backend: docker\n      env_passthrough:\n        - VANA_APP_KEY\n      docker_forward_env: # Docker backend only\n        - VANA_APP_KEY\n    ```\n\nStep 4 runs wherever the agent's shell commands run.\n\n## 4. Agent: identity\n\n```bash\nvana app register --app-url \u003Curl that identifies the agent>\nvana app whoami --json\n```\n\nUse the agent's homepage as the URL, such as\n`https:\u002F\u002Fgithub.com\u002Fopenclaw\u002Fopenclaw` or\n`https:\u002F\u002Fgithub.com\u002FNousResearch\u002Fhermes-agent`.\n\n`register` uses `VANA_APP_KEY` when it is set; otherwise it creates an\napp key separate from the owner's account, in the OS keychain where there\nis one, or in `~\u002F.vana\u002Fapp-key.json`. It is safe to run again. To run\ntwo agents on one machine, give each its own key through `VANA_APP_KEY`.\n`whoami` prints the **app address**; the owner needs it to fund escrow.\n\n## 5. Fund the agent's escrow (human step)\n\nThe simplest path for the person: run\n\n```bash\nvana app claim --json\n```\n\nand send them `data.claimUrl`. They open it while signed in to Vana\nAccount, click **Add and fund**, and land on the agent's Fund page, where\nthey deposit from a browser wallet: USDC.e on mainnet, faucet VANA on\nmoksha. The link is valid for about 50 minutes, and the first Account to\nopen it takes the app, so send it only to the owner. Run `claim` again for\na fresh link. Without the claim, the agent is in nobody's Apps list and\nAccount has no Fund page for it.\n\nFrom the agent's own terminal instead, the agent's key pays the gas, so\nits wallet needs VANA as well as the fee asset:\n\n```bash\nvana app escrow fund --amount 1 --json          # moksha: native VANA\nvana app escrow fund --amount 5 --asset \u003CUSDC.e token address> --network mainnet --yes --json\n```\n\nThe asset to fund is the one reads are priced in; a read without `--pay`\nnames it as `asset` in its exit-4 payload.\n\nThen check:\n\n```bash\nvana app escrow balance --json\n```\n\nFund small amounts. The agent can spend nothing beyond what escrow holds,\nand `--max-fee` caps each read.\n\n## 6. Grant: the agent asks, the owner approves\n\nAsk for the data the user chose. `request` does not check scope names: a\nwrong one still returns an approval URL, and fails only when the owner\nopens it. Get the exact names on the owner's machine, after `vana\nconnect` has collected the source:\n\n```bash\nvana data show github --json    # data.requestedScopes lists the scopes\n```\n\nThen the agent runs:\n\n```bash\nvana app request --scopes github.repositories --no-input --json\n```\n\nThat exits **7** with `approvalUrl`. Hand the URL to the owner. They sign\nin, review the scopes and approve only what the task needs. The agent then\npicks up the grant:\n\n```bash\nvana app requests show \u003Crequest-id> --json\n```\n\n`\u003Crequest-id>` is `data.requestId` from the request. `requests show`\nexits 0 either way: until the owner approves, `data.status` is `pending`\nand `data.grantId` is null. Once approved, `data.grantId` is the grant to\nread with. If the owner later revokes it, or a newer approval for the same\napp replaces its scopes, `data.live` turns false, `data.grant.state` says\n`revoked` or `replaced`, and `next` is a fresh `vana app request`.\n\nKeep requests to sources the Vana web app lists. A request that includes\na source it doesn't list cannot be approved on the web, and data from\nlocal-only sources (such as Claude Code history) can't be granted this way.\n\n## 7. The agent reads and pays\n\n```bash\nvana app read github.repositories --grant \u003Cgrant-id> --pay --max-fee 0.05 --json\n```\n\nWithout `--pay` it stops at exit **4** with the price. Every read is\ncharged, including a re-read of data that hasn't changed, so an agent that\nreads on a schedule spends on every run. The owner's own reads stay free.\n\n## 8. Review and revoke\n\nIn the Vana app, **Settings → Access history** lists each read the agent\nmade under its grant. **Revoke** ends its access: the Personal Server\nrefuses the next read and nothing is charged. To give access again, the\nagent asks again.\n\n## Networks\n\nEverything defaults to **mainnet**, where fees are real USDC.e.\nGrants, balances and server registrations do not carry across networks.\nTo try the whole flow on the testnet first, pass `--network moksha` on every\ncommand (every `vana` command accepts it), `vana server start` included: an\nagent reading on moksha cannot find a server registered only on mainnet.\n\n## Failure notes\n\n- Exit 5 on a read: the owner's Personal Server did not answer. Check\n  `vana server status` on the owner's machine. A read refused after a\n  revoke can also report exit 5 today; if the owner just revoked, treat it\n  as final and ask again rather than retrying.\n- Exit 4 with `--pay` set: `--max-fee` is below the price, or escrow is\n  empty. The limit is in the fee's own asset (USDC.e on mainnet).\n- Exit 7 from `request`: nobody has approved yet. Do not re-request in a\n  loop; poll `requests show` at a human pace.\n- The agent never runs `vana` commands: it did not load the builder\n  skill. Recheck step 2 for the agent the user picked.\n- `delivery: \"enclave\"` with `paid: false`: the owner serves this data\n  from a TEE sandbox, which the gateway currently admits at zero price.\n",{"name":5,"description":7},"main","a55d0a0006da6a743b6452796435afa6ed177dd8","skills\u002Fagent-access",0,"https:\u002F\u002Fgithub.com\u002Fvana-com\u002Fvana-cli","https:\u002F\u002Fgithub.com\u002Fvana-com","https:\u002F\u002Fvana.org","The first open protocol for data sovereignty.","2026-11-13 18:19:18",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add vana-com\u002Fvana-cli --skill agent-access",{"tool":39,"command":40},"bunx","bunx skills add vana-com\u002Fvana-cli --skill agent-access",{"tool":42,"command":43},"pnpm","pnpm dlx skills add vana-com\u002Fvana-cli --skill agent-access","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fvana-com\u002Fvana-cli\u002Fblob\u002Fmain\u002Fskills\u002Fagent-access\u002FSKILL.md",true,[],[48,55,62,69,76,83],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M7BD5YJ16KD0X3M379XA2A4A","verify-setup-resume","Verify BUI-936 setup\u002Fresume identities, funding readiness and overlapping resumes. Use when changing app request, test-credit transport or the request journal. Runs bounded TLA+ checks and offline real-code replays without live funding or paid reads.","https:\u002F\u002Fgithub.com\u002Fvana-com\u002Fvana-cli\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fverify-setup-resume\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":59,"categories":60,"keywords":61,"publishedAt":19},"01M7BD5V2155AM7W1PWRBY237Y","verify-server-supervision","Verify the local Personal Server stop and restart boundary. Use when changing the supervisor's backoff, port probing, replacement launch or Stop handling.","https:\u002F\u002Fgithub.com\u002Fvana-com\u002Fvana-cli\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fverify-server-supervision\u002FSKILL.md",[],[],{"skillNo":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"publishedAt":19},"01M7BD5PPVKFVVGS9B846B99ES","verify-paid-read-recovery","Verify interrupted and overlapping CLI paid reads, receipt-file writers and dead-owner recovery. Use when changing src\u002Fcli\u002Fapp\u002Fread.ts, src\u002Fcore\u002Freceipts.ts or src\u002Fcore\u002Freceipts-lock.ts. Runs TLA+ models and offline real-code replays without live paid reads.","https:\u002F\u002Fgithub.com\u002Fvana-com\u002Fvana-cli\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fverify-paid-read-recovery\u002FSKILL.md",[],[],{"skillNo":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"publishedAt":19},"01M7BD5JDEGV165NFZJWSWJBN4","verify-mainnet-registration","Verify mainnet app registration and Account association recovery. Use for the registration journal, gateway confirmation, retry, or ownership check.","https:\u002F\u002Fgithub.com\u002Fvana-com\u002Fvana-cli\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fverify-mainnet-registration\u002FSKILL.md",[],[],{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":19},"01M7BD5ETK33JAD1F1XQXGPYYG","verify-hosted-provisioning","Verify hosted Personal Server resource ownership across create and delete. Use for Connect's server API, provisioning-row updates, or deprovisioning admission.","https:\u002F\u002Fgithub.com\u002Fvana-com\u002Fvana-cli\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fverify-hosted-provisioning\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M7BD5BBKFHJ86JFBE8KT1S78","verify-export-settlement","Verify queued export owner binding and generation settlement. Use for result publication, pending upload receipts, or retry and sync changes.","https:\u002F\u002Fgithub.com\u002Fvana-com\u002Fvana-cli\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fverify-export-settlement\u002FSKILL.md",[],[]]