agentleFS
Sign inStart free
ogasurfproject-jpgogasurfproject-jpg/horizon-shield/skills/conduct-witness/SKILL.md

conduct-witness

Skill2 starsChanged 10 days agoMITView on GitHub

Walk an A2A or MCP agent as an independent witness and file the observation on the public conduct ledger under the user's name. Trigger with "be a witness for", "witness walk", "walk this agent", "証人になる", "この agent を歩いて", or when the user wants their own observation of an agent recorded where the agent cannot edit it.

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Do not walk an agent whose owner has declined measurement (listing: "decline" in the origin's /.well-known/mcp-conduct.json).

    line 14
  2. Very specific:

    If a fetch node shows http 403 at the edge, rerun with --transport curl.

    line 30

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Conduct witness

You are about to add one line to a record that the agent being walked did not write and cannot delete. Treat it that way: no guessing, no editing, no invented names.

Before running

  1. Ask the user once, if not already known in this conversation: their witness name (a person, a project, or "anonymous") and the vantage (where this machine is, in a few words: "laptop, Osaka", "VPS eu-west", "CI runner"). Never invent either. Reuse them for later walks in the same conversation.
  2. Confirm the origin is https:// and is the agent's card origin (the card lives at <origin>/.well-known/agent-card.json), not a page on the site.
  3. Do not walk an agent whose owner has declined measurement (listing: "decline" in the origin's /.well-known/mcp-conduct.json). If unsure, fetch that file first; absent means no objection.

Run

Preferred (uv installed):

uvx --from "git+https://github.com/ogasurfproject-jpg/horizon-shield#subdirectory=workers/hs-ledger/nenrin/a2a-conduct-walk" a2a-conduct-walk --origin <ORIGIN> --mode a2a --submit --witness-name "<NAME>" --vantage "<VANTAGE>"

Fallback (no uv):

curl -sSLO https://raw.githubusercontent.com/ogasurfproject-jpg/horizon-shield/main/workers/hs-ledger/nenrin/a2a-conduct-walk/a2a_conduct_walk.py && python3 a2a_conduct_walk.py --origin <ORIGIN> --mode a2a --submit --witness-name "<NAME>" --vantage "<VANTAGE>"

If a fetch node shows http 403 at the edge, rerun with --transport curl. If the user asked to look before filing, omit --submit first, show the assertions, then rerun with --submit only if they say so.

Report back, exactly

Give the user these five things from the output, verbatim: the purpose line (which endpoint), PASS n/5 or FAIL n/5, the record's sha256, the intake's answer (submitted ... http 200 or the reason it was not submitted), and the list of assertions with pass / FAIL / n/a. Say that a FAIL is filed the same way as a PASS and is a finding about the agent, not an error in the walk. Say where it will show up: the ledger bundles the pool once a day (00:30 UTC) at https://ledger.horizonshield.dev/ledger and the month's ring counts the witness by name.

Never

Never change the witness name to make a record look independent when it is not. If the user operates the agent being walked, put that in the name (for example "Example Corp (operator)"). Never submit twice to pad a count; one walk per agent per day is plenty. Never call this a rating, a score, or a verification of quality.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free