Introduce Tracy (🔎), a forensic code investigator who treats every
issue like a crime scene. Findings are graded as Confirmed, Deduced,
or Hypothesized, hypotheses are never deleted (only updated), and the
deliverable is a structured investigation file another engineer can
pick up cold.
Skills under src/bmm-skills/4-implementation/:
- bmad-agent-investigator: Tracy persona, registered in the module.yaml
agent roster (which surfaces her in Party Mode automatically) and in
module-help.csv. Menu codes IN (investigate) and CX (code archaeology).
- bmad-investigate: full case workflow with intake, scene survey,
evidence analysis, source-code trace, and finalization phases. Phase 4
folds archaeology techniques inline when a bug requires broader area
understanding. Resume mode is intrinsic auto-detection.
- bmad-code-archaeology: standalone code-area exploration when there is
no bug to chase.
Workflow integration:
- bmad-create-prd step-01-init now scans
{implementation_artifacts}/investigations/*.md and reports
investigationCount alongside briefs, research, brainstorming, and
project docs. The PM agent treats the case file as input context.
Greenfield/brownfield branching, the PRD template, and downstream
Phase 3/4 skills stay untouched. Restructuring step-02+ for fix-driven
mode is left for a maintainer-aligned PR.
- bmad-create-prd step-02-discovery echoes the new count.
Public docs:
- New explainer at docs/explanation/forensic-investigation.md covering
evidence grading, hypothesis discipline, the two downstream paths, and
the Tracy/Amelia separation.
- docs/explanation/named-agents.md adds Tracy and drops the hardcoded
count so the table is self-counting.
- docs/reference/agents.md adds Tracy with IN/CX triggers.
- docs/reference/workflow-map.md gains an Entry Points section
(greenfield vs working-backwards) and places bmad-investigate on the
Quick Flow track feeding bmad-quick-dev.
- The Quick Flow box in workflow-map-diagram.html now shows Tracy and
Amelia in series with an arrow between them.
* feat(core-skills): add bmad-customize for authoring _bmad/custom overrides
A conversational guide skill that helps users author or update TOML overrides
in _bmad/custom/ for customizable BMad agents and workflows. Covers per-skill
agent and workflow surfaces; central config is out of scope for v1.
- SKILL.md: six-step flow (intent, discover, route, compose, team-vs-user,
show-confirm-write-verify) with baked-in agent-vs-workflow routing heuristic
and a template-swap subroutine
- scripts/list_customizable_skills.py: stdlib-only scanner that enumerates
customizable skills across standard IDE install paths, reports surface type
and override status, PEP 723, 10 unit tests
- Reuses _bmad/scripts/resolve_customization.py for post-write verification
- Registered in core-skills/module-help.csv with menu code BC
* refactor(bmad-customize): apply QA pass (top 3 recommendations)
Applies the three highest-payoff themes from the quality analysis:
- Labeling + completion contracts: rename ## Purpose to ## Overview,
add domain framing (what customization means in BMad, typical user
arrival shapes), add an explicit Completion block with testable
conditions for "skill run is done"
- Hostile-environment robustness: add On-Activation preflight that
classifies no-BMad / BMad-without-resolver / full-install states,
instruct Step 2 to surface scanner errors[] and scanned_roots on
empty results, add resolver-missing fallback to Step 6.4, add a
re-enter-Step-4 recovery loop when verify shows the override didn't
take effect
- Returning-user and iteration experience: add "Audit / iterate"
intent class in Step 1, lead discovery with already-overridden
skills for that intent, read existing overrides in Step 3 before
composing, frame Step 4 as additive-on-top rather than fresh
authoring, give Cross-cutting intent an explicit Step 3 branch
that walks agent-vs-workflow with the user
Resolves 12 of 18 observations from the quality report. Lint clean
(scan-path-standards and scan-scripts both 0 findings). Unit tests
still 10/10.
* refactor(bmad-customize): derive skills root from install location
Previously the scanner hardcoded a list of IDE skill directories
(.claude/skills, .cursor/skills, .cline/skills, .continue/skills) and
scanned them relative to the project root. That was wrong: skills can
be installed either project-local or user-global, the IDE determines
the convention, and the set of valid locations is open-ended.
The scanner now derives its primary skills root from __file__ — the
running skill's own install directory is the authoritative location
for finding siblings. --skills-root overrides the default; --extra-root
(repeatable) adds additional locations for the rare mixed-install case.
Changes:
- list_customizable_skills.py: remove SKILL_ROOTS constant, add
default_skills_root() derived from __file__, rename scan_project
to scan_skills(skills_roots, project_root), add --skills-root and
--extra-root flags, de-dupe skills when the same name appears in
multiple roots (first wins)
- SKILL.md: update Step 2 to describe the scanner's derive-from-install
behavior and when to use --extra-root; drop the hardcoded IDE path
list from Notes
- tests: refactor setUp to place skills under a generic skills root
(not .claude/skills), add 3 new tests for multiple-roots merge,
duplicate-name precedence, and missing-root error reporting
* docs(customization): point users at bmad-customize as the guided path
Surface the new bmad-customize skill across the three customization
docs so users know they don't need to hand-author TOML to benefit
from the surface:
- customize-bmad.md: prominent tip at the top introducing the skill
as the guided authoring helper; updated the "Need to see what's
customizable?" troubleshooting tip to recommend the skill first
- expand-bmad-for-your-org.md: tip under prereqs noting every recipe
can be applied via the skill, with the recipes remaining the
reference for what to override
- named-agents.md: short paragraph in the customization section and a
link entry under the references list
Hand-authoring still works the same way; the skill is additive.
Central-config overrides are flagged as the current exception.
* docs(bmad-customize): steer users at bmad-builder instead of 'forking'
* fix(bmad-customize): reword description to pass file-ref validator
* refactor(bmad-customize): tighten description and expand module-help entry
- SKILL.md description: drop the catch-all 'or asks how to change the
behavior of a specific BMad skill' trigger clause that would fire in
casual discussion; keep the four explicit phrase triggers.
- module-help.csv: rewrite the description so bmad-help has real
routing material — names the concrete capabilities (persistent
facts, template swaps, activation hooks, menus), the scope routing,
and the value prop (no TOML hand-authoring). Matches the 'Use
when...' pattern other Core entries use.
* fix(module-help): quote bmad-customize description field that contains commas
* fix(bmad-customize): address PR #2289 review findings
- SKILL.md preflight: load root config from _bmad/config.toml and
config.user.toml (not .yaml) — the installer emits TOML; the YAML
references would have made the skill silently miss real user config
- SKILL.md resolver fallback (Step 6.4): read all three merge layers
when present (base / team / user) and describe the merge in
base → team → user order; the prior wording could describe the wrong
effective merge when the user wrote .user.toml on top of an existing
team .toml
- SKILL.md: replace bare 'docs/how-to/customize-bmad.md' references
(3 locations) with the public docs URL so users installing the skill
aren't pointed at a path they don't have locally
- list_customizable_skills.py: catch UnicodeDecodeError in
read_frontmatter_description so a non-UTF-8 SKILL.md can't abort
the whole scan
- list_customizable_skills.py: clarify exit-code contract in the
module docstring — errors[] is non-fatal by design, exit 2 is
reserved for invocation errors
- customize-bmad.md: tighten the tip to scope bmad-customize to the
per-skill surface; central-config is out of scope v1
- expand-bmad-for-your-org.md: same scoping — Recipes 1-4 can be
applied by the skill; Recipe 5 (central config) stays hand-authored
* fix(bmad-customize): markdownlint MD034 and validate-file-refs
- Wrap the three docs.bmad-method.org references as
[text](url) markdown links instead of bare URLs (MD034)
- Drop the {project-root}/ prefix on line 41's config.toml
references. validate-file-refs strips the template prefix and
tries to resolve 'config.toml' as 'src/config.toml'; sibling
skills (party-mode, retrospective, advanced-elicitation) all
reference '_bmad/config.toml' bare and pass CI — match that
pattern. The '(root level under {project-root}, installer-owned)'
parenthetical preserves the disambiguation.
* refactor(bmad-customize): cut token-wasting prose from SKILL.md
Down from 175 lines to 110. Removed:
- 'What customization means in BMad' architecture backgrounder — the
LLM reads the live customize.toml in Step 3; doesn't need the lore
- 'Desired Outcomes' section — retrospective narration of what the
6 steps already instruct
- 'Role' section — fluff; the flow itself defines the role
- 'Notes' section — sparse-override rule already in Step 4, IDE-path
note is commentary, docs link duplicates the out-of-scope section
- 'The scanner derives its skills directory from...' and 'returns JSON
with...' — commentary the LLM doesn't need; it runs the script and
sees the output
- 'that file IS the schema' and similar editorial asides throughout
- Explanatory clauses like 'silently drifts on every release' and
'trust the user's domain knowledge'
Kept everything that's load-bearing: preflight conditionals, intent
classification, routing heuristic, merge semantics, template-swap
subroutine, team-vs-user defaults, verify fallback and recovery loop,
completion conditions, out-of-scope list.
* refactor: remove bmad-skill-manifest yaml; introduce four-layer central config.toml
- Agent essence moves from per-skill bmad-skill-manifest.yaml files
into each module.yaml's `agents:` block (code, name, title, icon,
description). Per-agent customize.toml remains the deep-behavior
source of truth.
- Installer emits four TOML files:
_bmad/config.toml team install answers + agent roster
_bmad/config.user.toml user install answers
_bmad/custom/config.toml team overrides stub
_bmad/custom/config.user.toml personal overrides stub
Prompts declare scope: user to route answers to config.user.toml.
- resolve_config.py merges four layers: base-team -> base-user ->
custom-team -> custom-user.
- Three consumer skills (party-mode, advanced-elicitation,
retrospective) switched from agent-manifest.csv to the resolver.
- installer.js mergeModuleHelpCatalogs now takes the in-memory
agent list from ManifestGenerator -- no CSV roundtrip.
- Deleted: 6 bmad-skill-manifest.yaml files, agent-manifest.csv
emission, collectAgents/getAgentsFromDirRecursive,
paths.agentManifest().
* fix(installer): strip core-key pollution from [modules.*]; soften config headers
- writeCentralConfig now always strips core-module keys from every
[modules.<code>] bucket, even when the module's schema is not
available in src/ (external / marketplace modules like cis, bmb).
Core values belong in [core] only; workflows read them directly.
- When the module's own schema IS available (built-in modules),
also drop any key it does not declare as a prompt — same
spread-pollution filter as before, now layered on top.
- Section-aware headers on both _bmad/config.toml and
_bmad/config.user.toml: [core] / [modules.*] values are
editable (installer reads them as defaults on next install);
[agents.*] is regenerated from module.yaml and will be wiped —
overrides for agents go in _bmad/custom/config*.toml instead.
* docs: cover central config.toml + Diataxis prose pass across three files
Document the new four-file central configuration surface (_bmad/config.toml,
config.user.toml, and custom/ overrides) alongside the existing per-skill
customize.toml. Make editing rules, scope partitioning, and when-to-use-which
guidance explicit.
- customize-bmad.md: new "Central Configuration" section with editing rules,
three worked examples (rebrand, fictional agent, module settings override),
and a "when to use which surface" table. Converted five h4 headers to
bold paragraph intros per style guide.
- expand-bmad-for-your-org.md: two-layer mental model extended to three;
new Recipe 5 with three variants (rebrand, custom crew, pinned team
settings); reinforcement table extended.
- named-agents.md: noted the dual customization surface — per-skill shapes
behavior, central config shapes roster identity.
Diataxis prose pass applied across all three files: banned vocabulary
check, em-dash cap, hypophora / metanoia / amplificatio / stakes-inflation
cleanup, rhythm and burstiness fixes. Structural conformance verified;
markdownlint and prettier clean.
* test+docs: add central config unit tests; fix stale recipe count
- test: two new suites (35 + 36) covering writeCentralConfig and
ensureCustomConfigStubs. Verifies scope partitioning (user_name
lands only in config.user.toml), core-key pollution stripping
from [modules.*], unknown-schema fallthrough (external modules
survive without schema), agent roster baked into config.toml
[agents.*] only, stub-preservation on re-install. 44 new
assertions.
- docs: fixed four stale "four recipes" references to say "five"
after Recipe 5 (Customize the Agent Roster) was added. Touches
frontmatter, opening paragraph, Combining Recipes paragraph,
and the named-agents cross-link blurb.
* fix: address PR review feedback on central config
- resolve_config.py argparse: three-layer → four-layer description
- SKILL/workflow/explanation docs: document all four layers including
_bmad/config.user.toml (was missing from merge-stack descriptions)
- customize-bmad.md + installer headers: drop the false "direct edits to
config.toml persist" claim; installer reads from per-module config.yaml,
not central TOML, so direct edits get clobbered. Route users to
_bmad/custom/config.toml for durable overrides
- writeCentralConfig: warn loudly when a module.yaml can't be parsed
(previously silent — user-scoped keys could mis-file into team config)
- writeCentralConfig: preserve [agents.*] blocks for modules that didn't
contribute fresh agents this run (e.g. quickUpdate skipping modules
whose source is unavailable) so the roster doesn't silently shrink
- add extractAgentBlocks helper + Test Suite 37 covering preservation
Addresses comments from augmentcode and coderabbitai on PR #2285.
* feat(skills): TOML-based agent customization with stdlib Python resolver
Re-applies PR #2282's three-layer customization model (skill defaults →
team → user) but swaps YAML for TOML and uv for stdlib tomllib. Users
no longer need uv, pip, or a virtualenv — plain python3 (3.11+) is
sufficient, since tomllib shipped in the standard library.
## Schema changes vs PR #2282
- Flat agent schema: fields live directly under [agent], no nested
metadata/persona sub-tables. Easier to author, less indentation.
- Non-configurable identity: name and title are declared in
customize.toml as source-of-truth metadata (for future skill-manifest
generation) but SKILL.md ignores overrides there — identity is
hardcoded to preserve brand recognition.
- role redefined: now describes what the skill does for the user
within its module phase, not a restatement of the title.
- persistent_facts replaces the activation-time file-context load AND
the old memories concept. Entries can be literal sentences or
file: prefixed paths/globs; avoids collision with the upcoming
runtime memory sidecar.
- activation_steps_prepend / activation_steps_append harmonized across
agents and workflows (replaces agent-specific critical_actions).
- [workflow] namespace mirrors [agent] for workflow customization.
Same four structural rules, same field vocabulary.
## Resolver (src/scripts/resolve_customization.py)
Four purely structural merge rules, zero field-name hardcoding:
- Scalars: override wins
- Tables: deep merge
- Arrays of tables where every item has `code` or `id`: merge by
that key (matching keys replace, new keys append)
- Any other array: append
No removal mechanism — overrides cannot delete base items. Fork the
skill or override by code with a no-op value to suppress defaults.
## Agents ported (6)
All six BMad agents now ship customize.toml + rewritten SKILL.md:
analyst (Mary), tech-writer (Paige), pm (John), ux-designer (Sally),
architect (Winston), dev (Amelia). Each uses the same 8-step
activation template: resolve → execute prepend → adopt persona →
load persistent facts → load config → greet (with {agent.icon}) →
execute append → dispatch or present menu.
Step 8 supports fast-path invocation: "hey Mary, let's brainstorm"
dispatches the matching menu item directly after greeting, skipping
the menu render when intent is clear. Chat, clarifying questions,
and bmad-help remain available when nothing on the menu fits.
## Installer + tooling
- _bmad/scripts/ provisioned on install (copies src/scripts/)
- _bmad/custom/ seeded with .gitignore for *.user.toml on fresh install
- Non-module-dir filter extended to skip _memory, memory, docs,
scripts, and custom when scanning for modules
- Dead _config/agents/ directory no longer created
- metadata.capabilities removed from agent-manifest.csv and schema
- eslint config extended to cover src/scripts/**
- validate-file-refs.js knows about custom/ as install-only
## Deferred for follow-up
- bmad-product-brief workflow port (the pilot that demonstrates
[workflow] + on_complete)
- Translated docs (cs/fr/vi-vn/zh-cn) — regenerate from English
* feat(skills): port bmad-product-brief to TOML workflow customization
Completes the customization surface rollout by giving the product-brief
workflow the same override model as the six BMad agents, under the
[workflow] namespace instead of [agent].
## customize.toml
Mirrors the agent shape under [workflow] with:
- activation_steps_prepend / activation_steps_append (harmonized across
agents and workflows — same field names, same append semantics)
- persistent_facts with the file: convention, seeded with
file:{project-root}/**/project-context.md
- on_complete scalar (renamed from PR #2282's skill_end for clarity —
reads cleaner as "what runs when the workflow completes")
## SKILL.md
7-step workflow activation:
1. Resolve workflow block
2. Execute prepend steps
3. Load persistent facts (file: or literal)
4. Load config
5. Greet if not already
6. Execute append steps
7. Stage 1 — Understand Intent
python3 + stdlib tomllib invocation; no uv required.
## Prompt file changes
- Path normalization: ../agents/ → agents/, ../resources/ → resources/,
bare foo.md → prompts/foo.md. All references now resolve from the
skill root (matches the convention documented in SKILL.md).
- Paths: meta-line added to each of the 4 prompt files that reference
other files, reinforcing "bare paths resolve from skill root" so the
LLM doesn't lose the convention when operating two hops into a
prompt chain.
- finalize.md terminal stage now calls the resolver for
workflow.on_complete — non-empty values run as the final step.
## Validation
- Resolver output verified: 4 workflow fields returned cleanly.
- validate-file-refs.js: 254 files scanned, 139 refs checked, 0 broken.
- test:refs: passing.
* docs(skills): enterprise customization recipes + workflow template variable
Three independent improvements bundled because they share the same
surface (workflow/agent customization) and landed from the same design
discussion:
## Fallback sentence disambiguated (7 SKILL.md files)
The "if the script fails" fallback used to say `{project-root}/_bmad/
custom/{skill-name}.toml` for the team override and then just `{skill-
name}.user.toml` for the user override, leaving the user file's
location implicit. LLMs could reasonably guess skill root or project
root instead. Replaced with an unambiguous numbered list that spells
out the full path for every file in the merge chain.
## Product-brief: stage promotion + brief_template variable
- Promoted `## Stage 1: Understand Intent` from a nested step inside
"On Activation" to a top-level section. The previous "Step 7: Stage
1 — Understand Intent → Proceed to Stage 1 below" was mechanical
numbering pretending to be a step. Activation now ends cleanly at
Step 6; Stage 1 is a peer section.
- Added `brief_template` as a workflow-level scalar customization
defaulting to `resources/brief-template.md`. Stage 4 reads
`{workflow.brief_template}` instead of the hardcoded path, so orgs
can point at their own template under `{project-root}/...` without
forking the skill.
## New doc: docs/how-to/extend-bmad-for-your-org.md
Four worked recipes that together cover most enterprise scenarios:
1. Shape an agent across every workflow it dispatches (dev agent +
Context7 MCP + Linear search — the highest-leverage pattern)
2. Enforce org conventions inside a specific workflow (product-brief
+ compliance-field persistent_facts)
3. Publish completed outputs to external systems (product-brief +
Confluence + Jira via MCP, gated on user confirmation for Jira)
4. Swap in your own output template (product-brief + brief_template
variable swap)
Opens with the two-layer mental model (agent spans workflows,
workflow is local) so readers pick the right granularity before
reading any recipe. Closes with a "Combining Recipes" section
showing all four composed. Cross-linked from customize-bmad.md.
## Validation
- Resolver: workflow.brief_template returns the default cleanly.
- validate-file-refs.js: 254 files scanned, 146 refs checked
(+7 from this commit), 0 broken.
* docs(skills): encourage CLAUDE.md/AGENTS.md reinforcement of critical rules
Added a "Reinforce Global Rules in Your IDE's Session File" section to
extend-bmad-for-your-org.md. BMad customizations only load when a
skill activates, but IDE session files (CLAUDE.md, AGENTS.md, cursor
rules, copilot-instructions) load every turn — worth restating the
most critical rules there too so they survive ad-hoc chat outside a
BMad skill.
Includes a one-line example reinforcing the Recipe 1 Context7 rule,
plus a scope table that clarifies what each layer is for:
- IDE session file: universal, every session, keep succinct
- Agent customization: persona-specific, every dispatched workflow
- Workflow customization: one workflow run
Emphasizes brevity — noise in the session file crowds out signal.
* docs(skills): add Named Agents explanation doc
New docs/explanation/named-agents.md walking through the three-legged
stool (skills + named agents + customization) with the "Hey Mary,
let's brainstorm" activation flow as the narrative thread.
Covers:
- Why named agents vs menu-driven or prompt-driven alternatives
- The 8-step activation flow and what each step contributes
- How customization scales the model beyond a single developer
- Cross-links to the how-to docs for implementation details
Sits alongside brainstorming.md, quick-dev.md, party-mode.md in the
explanation folder — feature narratives for users who want to
understand why BMad is designed the way it is, not just how to use it.
* docs(skills): clarify that keyed-merge requires a single identifier key per array
Review feedback (PR #2284) flagged that the merge-rules wording was
ambiguous: "every item has a `code` or `id` field" could reasonably
be read as "each item individually has at least one of the two",
allowing arrays to mix `code` and `id` across items.
The resolver has always required all items share the *same* identifier
key (all `code`, or all `id`). Mixed arrays fall through to append —
intentional, because mixing identifier keys within one array is a
schema smell and any guess about which key should merge creates a
worse trap than the append-fallback.
Clarified in three places:
- Merge-rules table in customize-bmad.md: "every item shares the
**same** identifier field"
- `code`/`id` convention paragraph: "pick **one** convention ... and
stick with it across the whole array"
- Resolver docstring and `_detect_keyed_merge_field` docstring:
explicit note that mixed arrays fall through with rationale
No behavior change.
* docs(skills): address CodeRabbit review — fallback rules, OS claim, headless greeting
Three fixes from PR #2284 review feedback:
## 1. Fallback merge wording (7 SKILL.md files)
Every SKILL.md told the LLM to merge the three customization files
"in priority order (later wins)" when the resolver fails. That reads
as shallow last-write-wins — but the resolver does structural merge
(scalars override, tables deep-merge, code/id-keyed arrays merge by
key, other arrays append). Following the old wording manually would
have silently stripped base `principles`, `persistent_facts`, and
`menu` items whenever a team override was present.
Expanded the fallback sentence to restate the four structural rules
explicitly, matching the resolver's behavior.
Applied to all 6 agents + bmad-product-brief workflow.
## 2. Python 3.11 / OS shipping claim (customize-bmad.md)
The docs claimed "macOS 13+, Ubuntu 22.04+, Debian 12+, Fedora 37+
all ship 3.11 or newer." Inaccurate — Ubuntu 22.04 defaults `python3`
to 3.10.6 (3.11 is a separate package), and macOS doesn't really
ship Python by default anymore.
Replaced with honest guidance: check `python3 --version` and note
that macOS without Homebrew and Ubuntu 22.04 default to 3.10 or
earlier.
## 3. Autonomous mode greeting gate (bmad-product-brief)
Product-brief's activation-mode detection documents autonomous mode
as "produce complete brief without interaction" — but Step 5 greeted
unconditionally, adding conversational output before the headless
artifact. Gated the greeting on `{mode}` != `autonomous`.
## Dismissed (replied on thread)
- `.gitignore` migration from *.user.yaml to *.user.toml: YAML
installer code was in reverted #2282, never released. No users
affected. Same rationale as Augment's earlier thread.
Validated: 254 files, 146 refs, 0 broken. test:refs 7/7,
test:install 242/242.
* docs: rename Extend to Expand throughout customization docs
Anderewrey
Brian