diagnosing-bugs

Files (4)

• LICENSE
• SKILL.md
• openai.yaml · agents/
• hitl-loop.template.sh · scripts/

SKILL.md (first lines)

# Diagnosing Bugs

A discipline for hard bugs. Skip phases only when explicitly justified.

When exploring the codebase, read `GLOSSARY.md` (if it exists) to get a clear mental model of the relevant modules, and check ADRs in the area you're touching.

## Redact

This skill has you show commands, outputs and captured artifacts. **Redact every secret first**: write `<REDACTED>` in its place. Build loops against env vars, so the credential stays in the environment rather than in what you show. Captured artifacts carry auth headers: quote only the lines that carry the signal.

**Redact the whole tree, and treat listings as secret-bearing by default.** A filter that scrubs one level of a config misses a secret nested deeper, and process listings show argv, which carries tokens passed as flags or headers. Redact values at every depth, and show process or config listings only through a filter that rewrites secret values wherever they appear.

If the redacted output is not enough to diagnose the bug, say so and ask the user.

## Phase 1: Build a feedback loop

**This is the skill.** Everything else is mechanical. If you have a **tight** pass/fail signal for the bug (one that goes red on _this_ bug), you will find the cause; bisection, hypothesis-testing, and instrumentation all just consume it. If you don't have one, no amount of staring at code will save you.

Spend disproportionate effort here.

### Ways to construct one, in roughly this order

1. **Failing test** at whatever seam reaches the bug: unit, integration, e2e.
2. **Curl / HTTP script** against a running dev server.
3. **CLI invocation** with a fixture input, diffing stdout against a known-good snapshot.
4. **Headless browser script** (Playwright / Puppeteer) that drives the UI and asserts on DOM/console/network.
5. **Replay a captured trace.** Save a real network request / payload / event log to disk; replay it through the code path in isolation.
6. **Throwaway harness.** Spin up a minimal subset of the system (one service, mocked deps) that exercises the bug code path with a single function call.
7. **Property / fuzz loop.** If the bug is "sometimes wrong output", run 1000 random inputs and look for the failure mode.
8. **Bisection harness.** If the bug appeared between two known states (commit, dataset, version), automate "boot at state X, check, repeat" so you can `git bisect run` it.
9. **Differential loop.** Run the same input through old-version vs new-version (or two configs) and diff outputs.
10. **HITL bash script.** Last resort. If a human must click, drive _them_ with `scripts/hitl-loop.template.sh` so the loop is still structured. Captured output feeds back to you.

Build the right feedback loop, and the bug is 90% fixed.

### Tighten the loop

Treat the loop as a product. Once you have _a_ loop, **tighten** it:

- Can I make it faster? (Cache setup, skip unrelated init, narrow the test scope.)
- Can I make the signal sharper? (Assert on the specific symptom, not "didn't crash".)
- Can I make it more deterministic? (Pin time, seed RNG, isolate filesystem, freeze network.)

A 30-second flaky loop is barely better than no loop; a 2-second deterministic one is tight, a debugging superpower.

### Non-deterministic bugs

The goal is not a clean repro but a **higher reproduction rate**. Loop the trigger 100×, parallelise, add stress, narrow timing windows, inject sleeps. A 50%-flake bug is debuggable; 1% is not, so keep raising the rate until it's debuggable.

### When you genuinely cannot build a loop

Stop and say so explicitly. List what you tried. Ask the user for: (a) access to whatever environment reproduces it, (b) a redacted captured artifact (HAR file, log dump, core dump, screen recording with timestamps), or (c) permission to add temporary production instrumentation. Do **not** proceed to hypothesise without a loop.

### Delegating a fix

A subagent sent to fix a flaky, intermittent or one-off failure gets this skill named in its prompt and a stop condition: "if you cannot reproduce it in N attempts, report what you tried and change no production code." Without the loop rule in the prompt, a subagent treats plausible suspects as the cause and rewrites code to match them.

### Completion criterion: a tight loop that goes red

Phase 1 is done when the loop is **tight** and **red-capable**: you can name **one command** (a script path, a test invocation, a curl) that you have **already run at least once** (show the invocation and its output, redacted), and that is:

- [ ] **Red-capable**: it drives the actual bug code path and asserts the **user's exact symptom**, so it can go red on this bug and green once fixed. Not "runs without erroring"; it must be able to _catch this specific bug_.
- [ ] **Deterministic**: same verdict every run (flaky bugs: a pinned, high reproduction rate, per above).
- [ ] **Fast**: seconds, not minutes.
- [ ] **Agent-runnable**: you can run it unattended; a human in the loop only via `scripts/hitl-loop.template.sh`.

If you catch yourself reading code to build a theory before this command exists, **stop: jumping straight to a hypothesis is the exact failure this skill prevents.** No red-capable command, no Phase 2.

## Phase 2: Reproduce + minimise

Run the loop. Watch it go red as the bug appears.

Confirm:

- [ ] The loop produces the failure mode the **user** described, not a different failure that happens to be nearby. Wrong bug = wrong fix.
- [ ] The failure is reproducible across multiple runs (or, for non-deterministic bugs, reproducible at a high enough rate to debug against).
- [ ] You have captured the exact symptom (error message, wrong output, slow timing) so later phases can verify the fix actually addresses it.

### Minimise

Once it's red, shrink the repro to the **smallest scenario that still goes red**. Cut inputs, callers, config, data, and steps **one at a time**, re-running the loop after each cut, and keep only what's load-bearing for the failure.

Why bother: a minimal repro shrinks the hypothesis space in Phase 3 (fewer moving parts left to suspect) and becomes the clean regression test in Phase 5.

Done when **every remaining element is load-bearing**: removing any one of them makes the loop go green.

Do not proceed until you have reproduced **and** minimised.

## Phase 3: Hypothesise

Before ranking, measure the input the failing step actually consumed, through its own path or variable (size, hash, empty). A probe that rebuilds the input another way tests a different system.

Generate **3–5 ranked hypotheses** before testing any of them. Single-hypothesis generation anchors on the first plausible idea.

Each hypothesis must be **falsifiable**: state the prediction it makes.

> Format: "If <X> is the cause, then <changing Y> will make the bug disappear / <changing Z> will make it worse."

If you cannot state the prediction, the hypothesis is a vibe: discard or sharpen it.

**Show the ranked list to the user before testing.** They often have domain knowledge that re-ranks instantly ("we just deployed a change to #3"), or know hypotheses they've already ruled out. Cheap checkpoint, big time saver. Don't block on it; proceed with your ranking if the user is AFK.

## Phase 4: Instrument

Each probe must map to a specific prediction from Phase 3. **Change one variable at a time.**

Tool preference:

1. **Debugger / REPL inspection** if the env supports it. One breakpoint beats ten logs.
2. **Targeted logs** at the boundaries that distinguish hypotheses.
3. Never "log everything and grep".

**Tag every debug log** with a unique prefix, e.g. `[DEBUG-a4f2]`. Cleanup at the end becomes a single grep. Untagged logs survive; tagged logs die.

**Throwaway harnesses that bind-mount host filesystems isolate propagation at setup.** After any `mount --rbind` of an API filesystem (`/dev`, `/sys`, `/run`) into a chroot or repair root, immediately run `mount --make-rprivate <target>` before use. A recursive bind copies shared propagation, so unmounting the copy can unmount the original.

**Perf branch.** For performance regressions, logs are usually wrong. Instead: establish a baseline measurement (timing harness, `performance.now()`, profiler, query plan), then bisect. Measure first, fix second.

## Phase 5: Fix + regression test

From HexSleeves/skills, MIT.

More from HexSleeves/skills