Contributor Mode On

Contributor Mode On — opt in to widened /ievo:feedback diagnostics

Turns on a project-local consent flag, .ievo/contributor.flag, that widens what /ievo:feedback may OFFER to attach to a filed report. Off by default; disabled again with /ievo:contributor-mode-off.

Scope (read before enabling) — what this is, and is not

This is Phase 1 of a two-phase proposal (ievo-ai/skills#448). Only Phase 1 is built. Read both lists before enabling.

What turning this ON actually changes:

• /ievo:feedback gains one additional, always-optional attachment it may offer at report time: the existing scrubbed tool-failure/permission-denial capture stream — the same records /ievo:evo-auto-enable's signal: corrections+failures already accumulates under .ievo/evolution-candidates/, each one already redacted by scrub.mjs before it ever touched disk — redaction, not removal: a record still carries the failing call's own input arguments, so read Step 2's consent manifest below for exactly what that includes. Without this flag, /ievo:feedback never offers that stream, even if it has candidates.
• The environment context /ievo:feedback already collects for every report (client, versions, OS, project stack) is unaffected — that was already always included and this flag changes nothing about it.
• The existing per-report Submit / Cancel confirmation (/ievo:feedback Step 5) is completely unchanged: you still see the full title + body and approve before anything reaches a public issue, on every single report, contributor mode or not.

What this explicitly does NOT do (out of scope, not built here):

• No session-transcript export. Attaching a distilled session transcript/.jsonl export is "Phase 2" of #448 — a separate, larger, more security-sensitive capability with no design or approval yet. Turning this flag on attaches nothing transcript-shaped, ever.
• No standing consent to auto-post. A later #448 follow-up proposed an /ievo:i-am-contributor mode that would let /ievo:feedback skip the per-report confirmation entirely. That is a different, separate, explicitly out-of-scope proposal — this skill does not implement it, and enabling this flag never removes a confirmation step.
• Creates no new data by itself. If /ievo:evo-auto-enable's failure/denial capture was never turned on, or has no candidates yet, this flag changes nothing observable — /ievo:feedback still only offers the environment-context section it always has.

When to use

• User says "turn on contributor mode", "enable contributor mode", "I am a contributor", "I want to share more diagnostics", "widen feedback payload"
• User wants their /ievo:feedback reports to be able to include captured tool-failure/permission-denial records for maintainers to diagnose issues more precisely

Steps

1. Verify .ievo/ exists

If .ievo/ is absent → init hasn't been run in this project. Tell the user:

iEvo not initialized in this project. Run /ievo:init first.
Contributor mode widens /ievo:feedback's payload — nothing to widen yet.

Exit.

2. Show the consent manifest and confirm

This is the static, category-level consent manifest — shown in full before the flag is written, never skipped. Show the user this text (or render it verbatim in your own words, keeping every point):

Contributor mode — what turning this ON allows

When /ievo:feedback files a report while this is ON, it may OFFER (never
auto-attach) to include:

1. Full environment context (client, versions, OS, project stack) — already
   collected for every report, on or off. Unaffected by this flag.
2. The existing scrubbed tool-failure/permission-denial capture stream —
   only if /ievo:evo-auto-enable's "signal: corrections+failures" has
   separately been turned on and has candidates. Each record names the
   tool that failed or was denied, its error, and THAT CALL'S OWN INPUT
   ARGUMENTS — so a denied Write/Edit record carries the start of the
   file text that call was writing. scrub.mjs redacts secret-shaped
   values (provider tokens, NAME=value pairs), rewrites $HOME-absolute
   paths, and caps each record at 500 characters before it ever touches
   disk — it does NOT strip code or file contents. Read yours first:
   .ievo/evolution-candidates/*.jsonl. Nothing is attached without you
   choosing to, per report.

Every report still shows you the full title + body and asks Submit/Cancel
before anything reaches a public issue (github.com/ievo-ai/skills) — this
flag only widens what MAY be offered as an attachment choice. It never
auto-posts, and it does not attach a session transcript (a separate,
unbuilt, larger proposal).

Then ask via AskUserQuestion:

• Question: Turn on contributor mode for this project?
• Header: Contributor mode
• Options (single-select):
  • Enable (Recommended if you want to share more diagnostics) — description: Widens /ievo:feedback's optional attachments as described above. Reversible any time with /ievo:contributor-mode-off.
  • Cancel — description: Leave contributor mode off. /ievo:feedback keeps working exactly as it does today.

If the user picks Cancel, stop here — write nothing.

3. Write the flag file

If the user picked Enable, use the Write tool (NOT Bash) to create:

file_path: <project>/.ievo/contributor.flag
content:
  enabled: true
  enabled_at: <ISO-8601 UTC timestamp>
  enabled_by: <user identifier if known, else "user-invocation">
  scope: feedback-payload-widening

The file format is YAML for easy human reading, mirroring .ievo/debug.flag and .ievo/evo-auto.flag. scope: feedback-payload-widening names exactly what this flag does — so a future, separately-approved capability (e.g. a transcript-export phase) can add its own distinct scope value rather than silently expanding what an already-enabled flag implies.

4. Confirm to user

Print:

🧬 iEvo contributor mode ENABLED

Flag: .ievo/contributor.flag (commit to share the setting with teammates)

From now on, /ievo:feedback may offer to attach the scrubbed tool-failure/
permission-denial capture stream to a report, in addition to the environment
context it already always collects. Every report still requires your
Submit/Cancel confirmation before anything is posted.

Want that capture stream to actually have data to offer? Turn on
/ievo:evo-auto-enable's "signal: corrections+failures" option separately —
contributor mode only controls whether /ievo:feedback is ALLOWED to offer it,
not whether it's being captured.

Turn off: /ievo:contributor-mode-off

Rules

• Idempotent: if contributor mode is already on, just refresh enabled_at and confirm — never re-ask the consent question as if it were a fresh opt-in.
• Off by default, project-local: lives in .ievo/, not user config — per-project, and shared with teammates only if the flag file is committed.
• Never skip the consent manifest. Step 2's manifest text is shown in full on every fresh enable — this is the one place contributor mode's scope is stated before the user commits to it.
• Never widen beyond Phase 1. This skill enables ONLY the capture-stream attachment option in /ievo:feedback. It must never be extended to cover a transcript export or a standing auto-post consent without a fresh, separately-approved design — see ## Scope above.
• Creates no capture data itself. This flag is a permission gate on /ievo:feedback's attachment offer, not a capture mechanism — the underlying tool-failure/denial stream is /ievo:evo-auto-enable's signal: corrections+failures, opted into independently.

See also

• /ievo:contributor-mode-off — turn contributor mode back off
• /ievo:feedback — where the widened attachment option appears (Step 3.9)
• /ievo:evo-auto-enable — turns on the underlying tool-failure/permission- denial capture stream this flag can offer to attach
• plugins/ievo/scripts/scrub.mjs — the privacy scrub every captured record already passed through before it reached disk