Agents / examples
AGENTS.md1 starsChanged 58 days agoApache-2.0View on GitHub
One runnable offline snippet. It has no README on purpose, and no gate watching it. The smallest complete use of the harness as a library: load a config, run the engine against a null client, evaluate the gate. It exists to be copied into a user's own script, so it must stay short, offline and dependency-free. Nothing here ships, and nothing here is imported by the package. - The absence of a README.md is a decision. examples sits in the…
In this repository39 other files it gives its agents.
- agent-core/AGENTS.mdAGENTS.md
- .agents/AGENTS.mdAGENTS.md
- AGENTS.mdAGENTS.md
- behavioral-regression/AGENTS.mdAGENTS.md
- claude-foundation/AGENTS.mdAGENTS.md
- config/AGENTS.mdAGENTS.md
- corpora/AGENTS.mdAGENTS.md
- demo/AGENTS.mdAGENTS.md
- docs/AGENTS.mdAGENTS.md
- experiments/AGENTS.mdAGENTS.md
- flow-corpus/AGENTS.mdAGENTS.md
- flow-protocol/AGENTS.mdAGENTS.md
- openspec/AGENTS.mdAGENTS.md
- scripts/AGENTS.mdAGENTS.md
- skills/AGENTS.mdAGENTS.md
- tests/AGENTS.mdAGENTS.md
- claude-foundation/CLAUDE.mdCLAUDE.md
- merge-gate-auditorSkill
- update-executive-reportSkill
- update-executive-reportSkill
- architecture-drift-guardSkill
- commonSkill
- corpus-guardianSkill
- dataset-lintSkill
- deploySkill
- e2e-matrix-sentinelSkill
- eval-corpus-forgeSkill
- hierarchical-recursive-brainstormSkill
- model-benchSkill
- openai-judgeSkill
- openspec-implementation-reviewSkill
- openspec-peer-reviewSkill
- openspec-quality-planSkill
- pre-pr-gateSkill
- project-setupSkill
- quality-gateSkill
- refactoring-decomposerSkill
- repo-invariant-reviewSkill
- test-completeness-guardSkill
The file
Files
AGENTS.md — examples
One runnable offline snippet. It has no README on purpose, and no gate watching it.
The smallest complete use of the harness as a library: load a config, run the engine against a null client, evaluate the gate. It exists to be copied into a user's own script, so it must stay short, offline and dependency-free. Nothing here ships, and nothing here is imported by the package.
Map
| Path | Role |
run_offline.py | The whole example: load_config, EvalEngine.from_config, evaluate_gate |
datasets/sample.jsonl | The minimal dataset, and the precedent for harness-loadable data outside a package |
Diagram
flowchart LR
accTitle: what the offline example depends on
accDescr: The example loads the protected example config from the config directory and runs the engine against a null Langfuse client, so a change to that config breaks the example with no gate watching.
CFG["config/eval.example.yaml<br/>PROTECTED"]
subgraph EX["examples/ (you are here)"]
RUN["run_offline.py"]
DS["datasets/sample.jsonl"]
end
ENG["EvalEngine"]
NULL["NullLangfuseClient<br/>no network"]
GATE["evaluate_gate"]
CFG --> RUN
DS --> RUN
RUN --> ENG
ENG --> NULL
ENG --> GATE
classDef here fill:#e8f0fe,stroke:#1a73e8,stroke-width:2px
class EX hereRules that bite here
- The absence of a
README.mdis a decision.examplessits in the documented exempt set of the component-README job in.github/workflows/docs.yml, with the reason "runnable snippets, documented from the root README". Nothing is failing; do not add one to quiet a gate that is not firing. - This example reads a protected file it cannot protect.
run_offline.pyloads../config/eval.example.yaml. That config is in the protected set and this directory is not, so a reviewed change there can break this example and no reviewer is obliged to notice. Run it after any change to that config. - It must stay offline and import-clean.
NullLangfuseClientis the point: no SDK, no credential, no network. An example that needs a key is not an example. - Keep it one file. The value here is that a reader sees the whole flow at once. New
surface belongs in a test or in
../demo/, which already has an orchestrated script.
Verify
python examples/run_offline.pySubagents
| Task in this directory | Agent | Why |
| Check whether a renamed public symbol is still used here | explorer | Read-only; this file is outside the suite so a rename breaks it silently |
| Run the example after a config or engine change | test-runner | Has Bash; no gate runs this, so someone has to invoke it deliberately |
| Review an added dependency before it lands | narrow-critic | An import that needs a credential turns the example into a live call |
See also
| Doc | Read it when |
../config/README.md | You need to know what eval.example.yaml declares before changing it |
../config/AGENTS.md | You are editing the config this example loads and need the approval obligation |
../demo/AGENTS.md | You want an orchestrated multi-beat walkthrough rather than one snippet |
../corpora/README.md | You are adding data and need the versioned, manifested form of this precedent |
../src/eval_harness/README.md | You need the library surface this example is demonstrating |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

