writing a public participation policy for your agent

TL;DR

Publish a short page that says what your agent does in communities, where it participates, how often, how it discloses itself, and how to reach a human about it. Link it from the agent's profiles. A public policy turns "what is this bot doing here" from an accusation into a URL, and writing it forces you to decide the rules before an incident decides them for you.

writing a public participation policy for your agent

Use this when

• Your agent participates in any public community
• A moderator asks "what are the rules governing this account"
• You are briefing a fleet and want one canonical reference
• You need to show good faith to a skeptical venue
• You are formalizing norms you have been following informally

Not for this skill when

• You need a privacy policy or terms of service (legal documents, different skill)
• The policy is internal-only (then it is an ops runbook, not a public policy)
• You have no community participation yet (write it when the first venue goes live)
• You want the policy to substitute for good behavior (it documents behavior, it does not excuse bad behavior)

Steps

1. Write the five sections, plainly, on one page. Identity (what the agent is and who operates it), venues (where it participates), behavior (what it does and how often), disclosure (how it identifies itself), and contact (how to reach a human). One page, no legalese.

   # [Agent name] participation policy
   ## Identity
   ## Venues
   ## Behavior
   ## Disclosure
   ## Contact and complaints

Expected output: a one-page draft. Success check: a moderator understands the whole thing in three minutes.

1. Make the behavior section specific enough to be checkable. "Participates respectfully" is decoration; "replies at most 3 times per day per venue, mentions our tool only when directly relevant, never posts in announcement threads" is a policy. Specificity is what makes it falsifiable, and falsifiability is what makes it trustworthy.

   [ ] numeric caps stated
   [ ] self-mention rule stated
   [ ] named off-limits behaviors

Expected output: checkable commitments. Success check: a stranger could audit your last month against the policy.

1. Publish it at a stable URL and link it from every agent profile. The policy only works if moderators can find it in ten seconds from the account in question. Profiles link to the policy; the policy links back to the operator.

   published at: [url]
   linked from profiles: [list of accounts]

Expected output: a live URL, linked everywhere relevant. Success check: from any agent profile, the policy is one click away.

1. Add a changelog and update it when behavior changes. Policies rot like everything else; a dated changelog ("2026-09: lowered daily cap from 5 to 3") proves the policy describes reality. An undated policy describes aspirations.

   ## Changelog
   [date]: [what changed and why]

Expected output: a changelog with at least the initial entry. Success check: the latest behavior change appears in the changelog within a week.

1. Point to it instead of arguing. When questioned, the reply is "here is our policy: [link]; happy to discuss any part of it." The policy absorbs the interrogation so the conversation stays about specifics, not about your character.

   [ ] used the link in at least one real conversation

Expected output: the policy cited in practice. Success check: a questioning thread de-escalated after the link was shared.

Variant phrasings

Bot policy template

Template phrasing. Step 1's five sections are the template; copy and fill.

AI agent transparency page

Transparency phrasing. Same document, framed for public trust rather than moderator relations.

Should my bot have a public charter

Should-we phrasing. Yes, once it participates anywhere public; the cost is an afternoon, the value is every future "what is this" conversation.

Example participation policy for an agent

Example phrasing. The five sections filled with realistic values; adapt the numbers to your venues.

Why it happens

Suspicion thrives in ambiguity: an undisclosed, unexplained automated account invites the worst interpretation of every post. A public policy removes the ambiguity preemptively, which defuses most suspicion before it forms. It also disciplines the operator: writing "we do X" in public commits you to X in a way internal notes never do.

Edge cases / pitfalls

• A policy nobody follows is worse than no policy. It converts mistakes into lies; keep it accurate or take it down.
• Do not copy another project's policy verbatim. The numbers and venues must be yours, or the first audit exposes the fiction.
• The contact section needs a human who actually replies. A complaints inbox nobody reads is a trust incident waiting for its moment.
• Keep it short. Every section you add beyond the five is a section nobody will read and you will forget to update.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_uZlbMG81eJ2ajd3q4sSHXA