agento-report

name	agento_report
description	Generate a full agento PR status report — 6-point green checks, zero-touch rate, display inline, and post to Slack
type	skill

Purpose

Produce a comprehensive status report for all PRs agento is handling in jleechanorg/agent-orchestrator. Includes per-PR 6-green checks AND zero-touch-by-operator rate analysis. Display inline and post summary to Slack.

---

Execution Steps

Step 1 — Collect open PRs

Use REST (not GraphQL — GraphQL is frequently exhausted):

gh api "repos/jleechanorg/agent-orchestrator/pulls?state=open&per_page=30&sort=updated" \
  --jq '.[] | "\(.number)\t\(.head.ref)\t\(.title[:60])"'

Also collect recently merged (last 24h):

gh api "repos/jleechanorg/agent-orchestrator/pulls?state=closed&per_page=30&sort=updated&direction=desc" \
  --jq '.[] | select(.merged_at != null) | "\(.number)\t\(.head.ref)\t\(.title[:70])\t\(.merged_at)"'

Filter merged ones to last 24h by comparing .merged_at timestamp.

Step 2 — Per-PR data fetch (open PRs)

For each open PR number NUM, fetch mergeability, CI, and reviews via REST:

# Mergeability (REST returns boolean mergeable + string mergeable_state)
gh api "repos/jleechanorg/agent-orchestrator/pulls/NUM" \
  --jq '{mergeable, mergeable_state}'

# CI checks
SHA=$(gh api "repos/jleechanorg/agent-orchestrator/pulls/NUM" --jq '.head.sha')
gh api "repos/jleechanorg/agent-orchestrator/commits/$SHA/check-runs" \
  --jq '.check_runs[] | {name, status, conclusion}'

# Reviews
gh api "repos/jleechanorg/agent-orchestrator/pulls/NUM/reviews" \
  --jq '.[] | {user: .user.login, state}'

# Inline comments (check for High Severity / Critical / Major blockers)
gh api "repos/jleechanorg/agent-orchestrator/pulls/NUM/comments" \
  --jq '[.[] | select(.body | test("High Severity|Critical|Major"))] | length'

Step 3 — 6-point green check per open PR

Apply all 6 conditions:

#	Condition	Pass criteria
1	CI passing	All check-runs completed with SUCCESS, NEUTRAL, or SKIPPED — no FAILURE
2	No conflict	mergeable_state is clean (not dirty or unstable)
3	CodeRabbit APPROVED	Last coderabbitai[bot] review state is APPROVED
4	Bugbot OK	Last cursor[bot] check conclusion is neutral or success
5	No blocking comments	Zero inline comments with "High Severity", "Critical", or "Major"
6	Evidence PASS	/er comment with PASS exists (skip if no evidence bundle)

Status label (pick worst failing condition):

• GREEN — all 6 pass
• CONFLICT — mergeable_state is dirty
• CI_FAILED — any check has FAILURE conclusion
• CI_PENDING — checks still in_progress
• NO_CR — CodeRabbit hasn't APPROVED
• CR_CHANGES_REQUESTED — CodeRabbit posted CHANGES_REQUESTED
• COMMENTS — unresolved High/Critical/Major comments
• NO_EVIDENCE — evidence check failed (skip if N/A)

Step 4 — Zero-touch rate analysis (KEY ADDITION)

Definition (from ~/.openclaw/SOUL.md): Zero-touch-by-operator is measured by the [agento] prefix in the PR title. A merged PR with [agento] prefix means AO produced the work and brought it to merge without operator intervention.

For each merged PR in the window:

gh api "repos/jleechanorg/agent-orchestrator/pulls?state=closed&per_page=30&sort=updated&direction=desc" \
  --jq '.[] | select(.merged_at != null) | {
    number,
    title: .title[:70],
    merged_at,
    agento: (.title | test("^\\[agento\\]"))
  }'

Calculate:

• Total merged in window
• [agento] tagged (zero-touch)
• Non-[agento] (operator-assisted)
• Zero-touch rate = agento_count / total_merged * 100

For each non-[agento] PR, note WHY it wasn't zero-touch — common reasons:

• Missing [agento] prefix (worker didn't tag it)
• Operator had to fix lint/build errors on main
• Operator manually resolved merge conflicts
• Operator directly pushed code fixes
• CR review required operator comments

Step 5 — Format and display the report inline

## Agento Status Report — YYYY-MM-DD HH:MM

### Summary
- Repo: jleechanorg/agent-orchestrator
- Open PRs: N
- GREEN (ready to merge): N
- Not green: N
- Merged (last 24h): N
- Zero-touch rate (24h): X% (N/M [agento]-tagged of M total merged)

### Open PRs

| PR | Branch | Status | Blockers |
|----|--------|--------|----------|
| [#NUM](URL) title | branch | GREEN/status | blocker details |

### Merged (last 24h)

| PR | Title | Zero-touch? | Notes |
|----|-------|-------------|-------|
| [#NUM](URL) | title | [agento] YES / NO | reason if NO |

### Zero-Touch Analysis
- Rate: X% (N/M)
- Non-zero-touch PRs and why:
  - #NUM: reason (e.g., "operator fixed lint on main before CI could pass")
  - #NUM: reason (e.g., "worker didn't add [agento] prefix")

### AO Sessions
(tmux session count + key worker status)

Step 6 — Post Slack summary via MCP

Post to #ai-slack-test (channel ID: C0AKALZ4CKW):

mcp__slack__conversations_add_message(
  channel_id="C0AKALZ4CKW",
  text="*Agento Status Report — YYYY-MM-DD HH:MM*\n\nRepo: jleechanorg/agent-orchestrator\nOpen: N PRs | GREEN: N | Not green: N\nMerged (24h): N | Zero-touch rate: X%\n\n<per-PR details>"
)

---

Notes

• Scope: jleechanorg/agent-orchestrator (not jleechanclaw — that repo is deprecated for AO work).
• Use REST API (gh api) not GraphQL (gh pr view --json) — GraphQL is frequently exhausted.
• mergeable_state from REST: clean, dirty, unstable, unknown.
• Zero-touch convention: [agento] prefix in PR title (from ~/.openclaw/SOUL.md).
• If ao status is unavailable, use tmux list-sessions | grep -E 'ao-[0-9]+' as fallback.
• Always display inline report FIRST, then post Slack.
• The Slack post uses MCP, NOT curl.