All bugs start here. Skip phases only when explicitly justified.
Core methodology: Matt Pocock's 6-phase diagnose loop (Phase 1-6), with Phase 0 (context loading) and enhanced Phase 5-6.
Phase declaration rule: Every phase transition must be declared: ## Entering Phase N / ## Phase N complete. Never silently skip.
Phase 0 — Load context
Before touching code, read (skip if not present):
- CONTEXT.md — domain terms and concept relationships
- docs/specs/<area>.md — known pitfalls, invariants. Check if this bug matches an existing pitfall first.
- Relevant docs/adr/ — architectural decisions in the area
- Bug context — Issue full text, PRD + E2E report (if from E2E failure), or conversation context
If the Issue has a Debug Chain (see multi-session.md), continue from the last state — do not re-test ruled-out hypotheses.
Phase 1 — Build a feedback loop
This is the skill. Everything else is mechanical. If you have a fast, deterministic, agent-runnable pass/fail signal for the bug, you will find the cause — bisection, hypothesis-testing, and instrumentation all just consume that signal. If you don't have one, no amount of staring at code will save you.
Spend disproportionate effort here. Be aggressive. Be creative. Refuse to give up.
Ways to construct one — try them in roughly this order
- Failing test at whatever seam reaches the bug — unit, integration, e2e.
- Curl / HTTP script against a running dev server.
- CLI invocation with a fixture input, diffing stdout against a known-good snapshot.
- Headless browser script (Playwright / Puppeteer) — drives the UI, asserts on DOM/console/network.
- Replay a captured trace. Save a real network request / payload / event log to disk; replay it through the code path in isolation.
- 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.
- Property / fuzz loop. If the bug is "sometimes wrong output", run 1000 random inputs and look for the failure mode.
- 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 runit. - Differential loop. Run the same input through old-version vs new-version (or two configs) and diff outputs.
- HITL bash script. Last resort. If a human must click, drive them with
scripts/hitl-loop.template.shso the loop is still structured. Captured output feeds back to you.
Build the right feedback loop, and the bug is 90% fixed.
Iterate on the loop itself
Treat the loop as a product. Once you have a loop, ask:
- 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 loop is a debugging superpower.
Non-deterministic bugs
The goal is not a clean repro but a higher reproduction rate. Loop the trigger 100x, parallelise, add stress, narrow timing windows, inject sleeps. A 50%-flake bug is debuggable; 1% is not — 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 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.
Do not proceed to Phase 2 until you have a loop you believe in.
Phase 2 — Reproduce
Run the loop. Watch the bug appear.
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.
Do not proceed until you reproduce the bug.
Phase 3 — Hypothesise
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.
Root-cause depth: Your hypothesis should explain WHY the bug occurs, not just WHAT occurs. "User object is null" is a symptom. "User query returns null because the session cache expires before the token does" is a root cause. Keep asking "why?" until you reach a cause that, if fixed, eliminates the symptom without special-casing.
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.
If uncertain about fix approach for multi-module / state-machine / migration bugs, use /prototype to validate before modifying production code.
Phase 4 — Instrument
Each probe must map to a specific prediction from Phase 3. Change one variable at a time.
Tool preference:
- Debugger / REPL inspection if the env supports it. One breakpoint beats ten logs.
- Targeted logs at the boundaries that distinguish hypotheses.
- 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.
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
Fix anti-patterns (read before writing any fix code)
- Fixing at the symptom site: If your fix is a null check, try-catch, or fallback at the crash point, you're hiding the bug, not fixing it. Trace back to where the incorrect state was created.
- Fixing by addition only: Bug fixes should often REMOVE or CORRECT code, not add new code. If your fix is purely additive (new if-branch, new wrapper, new special case), ask whether existing code is wrong rather than missing.
- Fixing by loosening: If your fix makes validation more permissive, accepts wider input, or skips a check — you're probably creating a new bug to hide the current one.
- Fixing by duplication: If your fix copies existing logic to handle the edge case separately, you now have two places to maintain. Extend the original logic instead.
- Fixing with unexplained magic: If your fix works but you can't explain the causal chain from fix → symptom disappears, you haven't found the root cause. A correct fix has an obvious, explainable connection to the hypothesis from Phase 3.
Regression test
Write the regression test before the fix — but only if there is a correct seam for it.
Do not write tests that merely restate the implementation. A test that mocks all dependencies and then asserts the mocks were called provides zero confidence — it will pass even if the implementation is completely wrong. Tests must verify observable behavior through real (or realistically integrated) code paths.
A correct seam is one where the test exercises the real bug pattern as it occurs at the call site. If the only available seam is too shallow (single-caller test when the bug needs multiple callers, unit test that can't replicate the chain that triggered the bug), a regression test there gives false confidence.
If no correct seam exists, that itself is the finding. Note it. The codebase architecture is preventing the bug from being locked down. Flag this for Phase 6.
If a correct seam exists:
- Turn the minimised repro into a failing test at that seam.
- Watch it fail.
- Apply the fix.
- Watch it pass.
- Re-run the Phase 1 feedback loop against the original (un-minimised) scenario.
Phase 6 — Cleanup + knowledge capture (NEVER SKIP)
Cleanup checklist:
- Original repro no longer reproduces (re-run Phase 1 loop)
- Regression test passes (or absence of seam documented)
- All
[DEBUG-...]instrumentation removed - The correct hypothesis is stated in the commit message
Knowledge capture — answer all three, even if the answer is "none":
Q1: "What domain knowledge did I learn that I didn't know before?"
→ update CONTEXT.md (new terms, concept relationships, corrections). Format: follow CONTEXT-FORMAT.md.
Q2: "What pitfalls should the next person writing code here know?"
→ write/update docs/specs/<area>.md. Format: [condition] → [invariant] + [consequence of violation].
- Example: "When modifying SubscriptionCredit.expiresAt, must sync tokenRemainQuota on the gateway side. Violation: quota mismatch between gateway and billing."
- Example: "Trial credit cleanup only triggers on first payment, not on renewal. Violation: user's historical quota zeroed on renewal."
Q3: "What architectural change would fundamentally prevent this class of bug?"
→ hand off to /improve-codebase-architecture with specifics, or record "none".
Path escalation: design flaw (not bug) → /grill-with-docs → /to-prd. Scope exceeds Issue → file a new Issue.
Exit routing: if entered from E2E failure → re-run /e2e-verify after fix. If entered from /functional-test → return and continue remaining tests.
Methodology flaw noticed? Tell user directly, or append to bugfix/IMPROVEMENTS.md if autonomous.
See multi-session.md for handoff protocol and debug chain format. See handoffs.md for full entry/exit point reference.