Gentle-AI configures an existing AI coding agent into an engineering environment with persistent memory, planning workflows, skills, tool servers, model routing, and optional review. Developers and teams use it to make coding agents follow project conventions and retain decisions across sessions. The catalogue entries are its skills, commands, agents, and instruction.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add Gentleman-Programming/gentle-ai --skill issue-root-resolutiongit clone --depth 1 https://github.com/Gentleman-Programming/gentle-aiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/gentleman-programming/gentle-ai/issue-root-resolution)<a href="https://agentmods.dev/skills/gentleman-programming/gentle-ai/issue-root-resolution"><img src="https://agentmods.dev/badge/skills/gentleman-programming/gentle-ai/issue-root-resolution.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00045 | $0.01108 |
| Opus 5 | $0.00023 | $0.00554 |
| Sonnet 5 | $0.00009 | $0.00222 |
| Haiku 4.5 | $0.00005 | $0.00111 |
Grade A, and why
issue-root-resolution scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 56 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Activation Contract
Load when auditing a defect backlog for shared root causes, proposing a fix for an issue cluster, or closing issues as resolved/outdated. Complements rdd-defect-workflow (single-defect flow) — this skill governs the cluster-level method.
Hard Rules
- Read full bodies and comments, never titles. Classify against the actual code or PR diff, not descriptions of it.
- Verify every "fixed" or "broken" claim against current
origin/mainthe day you act. Compare report dates to fix merge dates: a repro filed before the fix landed is evidence about old builds, not current code. - Cluster defects by causal root, not surface. Each root gets a measured row in the meta-issue (#2471 style): issues attached, fix shape, state.
- Before proposing any fix, produce a read-only mechanism map: one file:line anchor per claim, mechanism explained as implemented today. Report claim-vs-code mismatches explicitly; never force evidence to fit the hypothesis. Let the map shrink the proposal.
- Rank solutions by what they DELETE: (1) removes a mechanism so the class becomes impossible, (2) static guard-ratchet making reintroduction a test failure, (3) localized predicate fix behind a failing repro test. New surface (verbs, flags, mechanisms) is last resort — defer until re-verified evidence demands it.
- Extract maintainer choices as named D-items with recommended defaults. A maintainer condition recorded in an issue thread outranks any plan table, including yours.
- Close only with evidence, one rule per closure: (A) fixed on main, cite commit AND proving test; (B) superseded by recorded maintainer decision; (C) surface no longer exists; (D) duplicate of a fixed issue. Comment before closing: what resolved it, verified today, reopen invitation. When in doubt, do not close — list as borderline.
- Clean break, never compatibility. When a format or identity changes, bump its versioned tag as a REPLACEMENT: old records become outdated and fail closed with an actionable refusal naming the rerun. Never write dual recognition, legacy fallbacks, or compat shims; that residue is what the deletion criterion exists to prevent. Stored bytes are never rewritten and stay readable for forensics, but carry no gate or lifecycle validity.
- Implementation follows the waves recipe: repro as failing test first, byte-stable goldens as defect signal, independently revertible slices, one writer per slice.
- A universal guard's verification matrix is
go test ./...at the repository root, never a curated package list. A guard that forbids a shape breaks every fixture that relied on it, including in packages the change never touched. - One worktree per writer, always, including the orchestrator. Before editing inline, check whether a delegated worker holds that path; if so, create a separate worktree rather than reusing it.
- Audit every worker report yourself: re-run its key verification, decoy-test any new guard, spot-check its diffs. A self-report is a claim, not evidence.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 56 lines · 45 tokens per session scan A 9236a56a8eec
issue-root-resolution is a skill published in the GitHub repository Gentleman-Programming/gentle-ai (6,328 stars, last pushed today), licensed MIT. It adds 45 tokens to every session and 1,108 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
incident-response
Incident management lifecycle — triage, communicate, mitigate, postmortem. Three modes — new (start incident), update (status update), postmortem (blameless RCA report).
discover
Primary router for discovery, debugging, investigation, quality, and exploration workflows. Analyzes user intent and dispatches to debug, bughunt, quick-review, deep-review, coverage, testing-strategy, learn, tour, impact, assist-review. The recommended entry point for any 'find out', 'check', 'review', or…
repo-consistency-sweep
Proactive defect-class detection that handles the lower-value half of code review (per Bacchelli and Bird 2013) so human reviewers stay focused on design, intent, and knowledge transfer. Catches convention drift, ordering bugs, type-safety gaps, security and multi-tenant invariants (CWE-grounded), and operability…
post-mortem
Diagnose instruction defects and optionally submit Rosetta GitHub issue.
memorix-troubleshooting
Use when Memorix MCP, setup, project binding, HTTP control plane, hooks, skills, or agent integration is missing, stale, or failing.
ijfw-review
Use when the user asks for a review of any artifact -- code diff, PR, book chapter, campaign brief, landing-page copy, or design tokens. Trigger: review, code review, review this, PR review, review my X, review chapter, review brief, review page, /ijfw-review.