Passed 17 of 21 test cases with the skill, and 7 of 21 without it.
Without skill33%
With skill79%
Checks it passed that the model alone missed
Categorizes security concerns under the SEC issue label.
Evaluates context map staleness using a 30-day threshold.
Invokes agent-ops-interview for each unconfirmed section gap.
and 17 more in the full report
12 fixed7 passed either way4 still failing1 broke
Pass-rate lift
+46 pp
Tokens
+47%
Agent turns
0%
Response time
-51%
Model gemini-3.6-flashMeasured 8/18/2026v119 of 24 cases passed with skill · 21 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 12 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 12
Prompt
While inspecting the codebase during constitution initialization, we noticed missing environment configuration templates and an outdated CI workflow file. Which issue discovery category applies to missing configurations or environment issues?
What was checked, without and with the skill
Checked
Without skill
With skill
Categorizes missing configurations and environment issues under the CHORE label.
Missed
Passed
case-18 · gemini-3.6-flash · 8/18/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md167 linessingle file
Constitution workflow (mandatory before baseline)
Goal
Make .agent/constitution.md baseline-ready by confirming:
allowed/restricted/forbidden work areas
single-line build/lint/test/format commands
environment assumptions
project-specific constraints
issue-first workflow expectations
Pre-requisite: Context Map
Before starting constitution interview, ensure a context map exists:
Check if .agent/map.md exists
If NO map exists:
📍 No context map found. Creating one first to inform constitution questions...
→ Invoke agent-ops-context-map to generate .agent/map.md
If map exists but is stale (>30 days old or significant changes detected):
📍 Context map may be outdated. Refresh? [Y]es / [N]o
Use the map to:
Pre-populate candidate work areas
Identify build/test tooling
Understand project structure for interview questions
Procedure
Ensure context map exists (see above)
Run tool detection — invoke agent-ops-tools or aoc tools scan --save
Creates .agent/tools.json with available development tools
Identifies missing recommended tools based on detected project type
Populates "Available tools" section in constitution
Inspect repo evidence (README, CI workflows, package/build files). Do not guess.
Draft v0 constitution:
every inferred item must cite its evidence ("CANDIDATE from ")
anything without evidence is TODO + UNCONFIRMED
use detected tools to suggest build/lint/test commands
Detect available development tools at session start. Saves to .agent/tools.json and warns about missing required tools. Works with or without aoc CLI installed.