Borrowing it
Nothing to install: this file belongs to Nam-Cheol/namba-ai. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/Nam-Cheol/namba-ai/main/.agents/skills/namba-harness/SKILL.mdgit clone --depth 1 https://github.com/Nam-Cheol/namba-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/nam-cheol/namba-ai/namba-harness)<a href="https://agentmods.dev/skills/nam-cheol/namba-ai/namba-harness"><img src="https://agentmods.dev/badge/skills/nam-cheol/namba-ai/namba-harness/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/nam-cheol/namba-ai/namba-harness"><img src="https://agentmods.dev/badge/skills/nam-cheol/namba-ai/namba-harness.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00018 | $0.00858 |
| Opus 5 | $0.00009 | $0.00429 |
| Sonnet 5 | $0.00004 | $0.00172 |
| Haiku 4.5 | $0.00002 | $0.00086 |
Grade A, and why
namba-harness 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 11d 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 — 36 lines — stays where its author put it; the contents beside it link to each section on GitHub.
State effect: mutating workflow entry point. Use help/probe paths read-only, and otherwise expect repository state or GitHub state to change.
Generated instruction contract for this command skill:
- Purpose: keep the role or command scope explicit, bounded, and testable.
- Boundary: honor read-only versus mutating state effects, configured sandbox mode, and assigned file or workflow ownership.
- Required output: report concrete actions, changed paths or artifacts, validation evidence, and pass/fail status or blockers.
- Pass/fail criteria: claim success only when acceptance criteria and configured validation are satisfied; otherwise name the exact blocker and impact.
- Evidence expectations: cite source artifacts such as SPEC files,
.namba/configs, diffs, test output, PR/check links, or generated manifests instead of relying on unsupported assertions. - Security responsibilities: never expose or commit secrets; treat auth, privacy, destructive commands, permission changes, and external network or credential use as security-sensitive.
- Destructive command and escalation policy: do not run destructive commands unless explicitly requested; request approval for privileged, networked, or sandbox-blocked actions only when the active approval mode allows it, and otherwise report the blocker or use a safe non-escalating path.
- Fallback implementer boundary: if a specialist path is unavailable and the main/default implementer takes over, stay within the assigned scope and preserve the same evidence and validation duties.
- Portability: keep durable guidance non-project-specific unless the current repository config or SPEC explicitly provides the project detail.
Use this skill when the user explicitly says $namba-harness, namba harness, or asks to create a harness-oriented SPEC package.
Behavior:
- Before reading project docs, checking Git state, or running the CLI, run the same Namba clarification gate used by
namba planon the user's raw request. - If the request is short, broad, or missing target surface, user flow, scope boundaries, constraints, acceptance criteria, or validation, do not run
namba harnessyet. Ask 1-3 concise questions first. - Treat Codex Plan mode as the preferred clarification surface, then pass only the refined Goal/Scope/Constraints/Acceptance description to
namba harness "<refined description>". - Prefer the installed
namba harnessCLI when available. - Use this path for reusable agent, skill, workflow, orchestration, or evaluation scaffolding when the user wants a reviewable SPEC first instead of generating the repo-local skill or agent artifact directly through
$namba-create. - Start with the same dedicated-branch planning contract as
namba plan, and use--current-workspaceonly when the user intentionally wants to scaffold on the current branch without creating a dedicated SPEC branch. - Do not create planning worktrees here either; temporary worktrees belong to overlapping
namba run SPEC-XXX --parallelexecution only. - Create the next sequential
SPEC-XXXpackage under.namba/specs/without inventing a second artifact model. - Seed
.namba/specs/<SPEC>/reviews/with product, engineering, design, and aggregate readiness artifacts so the review flow stays aligned withnamba plan. - Keep command-entry skill guidance lean; move long PR-thread, CI-log, frontend, or MCP recipes into existing references or deterministic helper-script candidates instead of creating new standalone skills.
- Evaluate deterministic helper-script candidates before implementation: they need
--help, read-only defaults, bounded output, explicit network/auth assumptions, fixture or local-server tests, and no destructive or third-party app coupling. - For large managed-skill changes, inventory affected source and generated surfaces, classify mechanical versus behavioral edits, update templates first, regenerate, review generated diffs, and validate.
- For harness/MCP quality, prefer workflow-first designs over raw endpoint wrappers, require context-budgeted outputs with pagination or truncation expectations, produce actionable errors, and define evaluation scenarios that are independent, read-only, realistic, verifiable, and stable.
- Keep the output Codex-native and avoid Claude-only runtime primitives in the planned contract.
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.
- 11d ago First seen · 36 lines · 18 tokens per session scan A a882ae7faa30
namba-harness is a skill published in the GitHub repository Nam-Cheol/namba-ai (11 stars, last pushed 19d ago), licensed MIT. It adds 18 tokens to every session and 858 once invoked, about $0.0001 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 2026-08-30.
Other skills, from other repositories
agf
Find local coding-agent sessions, inspect bounded metadata or summaries, and prepare a read-only resume plan with AGF. Use for finding prior work across agent CLIs, not for executing agents, deleting sessions, or reading full transcripts.
dynamic-resources
Example skill loaded from resourcesdiscover.
repo-hygiene
Use when the scheduled repo-hygiene workflow runs from GitHub Actions (or an operator dry-run) to scan the repository for small, certain docs/test/code hygiene issues and fix them as one batched branch.
find-simplifications
Use for a periodic repo-wide sweep of qwen-code for accumulated excess surface — dead components and files, orphaned locale keys, exports nothing consumes, added-then-removed scaffolding — filing candidates on a tracking issue and landing only what a maintainer has said yes to. Repo-wide and evidence-first; every…
swarm-pr-review
Run a graph-guided, tool-augmented PR review using context packing, parallel exploration, mandatory repository-agnostic risk-family coverage with dispatch scaled to diff size and risk, independent reviewer validation, critic challenge, and metrics writeback. Use for deep pull request review with low false-positive…
bundle-safety
Bundle transform safety — minification variant selection, consumer-constraint verification, identifier preservation, and namespace re-export coverage for build output.