Add opt-in strict mode that fails creation if required sections are
missing. Uses ValidateTemplate() under the hood.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Checks issues for missing recommended sections based on type:
- bd lint # Lint all open issues
- bd lint bd-xxx # Lint specific issue
- bd lint --type bug # Filter by type
- bd lint --json # JSON output for CI
Supports both daemon and direct mode. Exit code 1 when warnings found.
Part of opt-in validation epic (bd-ou35).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The close command now properly routes to different beads directories
based on issue ID prefix, matching the behavior of bd show/update.
Changes:
- Check needsRouting() for each ID in both daemon and direct mode
- Handle routed IDs via resolveAndGetIssueWithRouting()
- Close issues in the correct remote store
Fixes bd-3jrb
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The bd update command now checks needsRouting() before attempting daemon
RPC resolution. When an issue ID (like hq-eggh5) routes to a different
beads directory, the update bypasses the daemon and uses direct mode with
the routed storage.
This enables polecats in gastown to update HQ beads (hq-* prefix) and
vice versa. The fix mirrors the routing pattern already used by bd show.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The ValidateAgentID function was hardcoded to only accept 'gt-' prefix,
but beads rig uses 'bd-' prefix. Now accepts any valid prefix.
Changes:
- Extract prefix dynamically instead of hardcoding 'gt-'
- Accept IDs like bd-mayor, bd-beads-polecat-pearl
- Update error messages to be prefix-agnostic
- Add test cases for alternative prefixes
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Adds --pretty flag to bd ready for consistency with bd list.
Uses the same displayPrettyList() function to show issues in
a tree format with status/priority symbols.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove bd_new (31MB binary that should never have been tracked)
- Remove @AGENTS.md (duplicate that was already removed but still tracked)
- state.json already in .gitignore and not tracked
- history/EPHEMERAL_MOLECULES_DESIGN.md does not exist
(bd-ov1u)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The bd update command now checks needsRouting() before attempting daemon
RPC resolution. When an issue ID (like hq-eggh5) routes to a different
beads directory, the update bypasses the daemon and uses direct mode with
the routed storage.
This enables polecats in gastown to update HQ beads (hq-* prefix) and
vice versa. The fix mirrors the routing pattern already used by bd show.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The ValidateAgentID function was hardcoded to only accept 'gt-' prefix,
but beads rig uses 'bd-' prefix. Now accepts any valid prefix.
Changes:
- Extract prefix dynamically instead of hardcoding 'gt-'
- Accept IDs like bd-mayor, bd-beads-polecat-pearl
- Update error messages to be prefix-agnostic
- Add test cases for alternative prefixes
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Two fixes for bare repo worktree setups:
1. fork_protection.go: Use git.GetGitDir() instead of hardcoding .git
In worktrees, .git is a file containing gitdir path, not a directory.
Using GetGitDir() handles this correctly.
2. sync_git.go: Simplify gitHasBeadsChanges to use absolute paths
The previous code used git -C main-repo-root status, but in bare
repo worktrees GetMainRepoRoot() returns the parent of the bare repo
which is not a valid working tree. Using absolute paths without -C
lets git find the repo from cwd, which always works.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The memory storage GetDependencyTree was missing the root node at depth 0,
causing --no-db mode to only show dependencies without the root issue.
Fixed by:
- Adding root node at depth 0 before listing dependencies
- Setting ParentID on child nodes for proper tree rendering
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The last-touched file was accidentally removed from GitignoreTemplate
in commit 310d374 (GH#814 fix). This restores it and adds it to
requiredPatterns so bd doctor will detect outdated gitignores.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add --notes flag to bd create command, enabling agents to set notes
during issue creation instead of requiring a separate update command.
Motivation: AI agents repeatedly tried to use --notes with bd create.
Context is fresh at creation time - forcing a two-step process means
context loss or workflow interruption.
Changes:
- cmd/bd/flags.go: Added --notes flag to common issue flags
- cmd/bd/create.go: Read and pass notes in both RPC and direct modes
- cmd/bd/update.go: Removed duplicate --notes flag definition
- internal/rpc/protocol.go: Added Notes field to CreateArgs
- internal/rpc/server_issues_epics.go: Process Notes in handleCreate
- cmd/bd/create_notes_test.go: Comprehensive test coverage
- website/docs/cli-reference/issues.md: Documentation
Also adds gitignore entries for Augment AI and .beads/redirect.
Co-authored-by: Leon Letto <lettol@vmware.com>
The fork protection logic incorrectly treated all repos where
origin != steveyegge/beads as forks, including user's own projects
that just use beads as a tool.
Changes:
- Add isForkOfBeads() that scans ALL remotes for steveyegge/beads
- Only apply protection when a beads-related remote exists
- Add git config opt-out: `git config beads.fork-protection false`
(per-clone, never tracked, matches beads.role pattern)
Test coverage for 8 scenarios plus edge cases for config values.
Adds #nosec G204 annotations to exec.Command calls that use
hardcoded paths. Fixes pre-existing lint failure exposed by PR #832.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Executed-By: beads/crew/dave
Rig: beads
Role: crew
* test(daemon): Prove autoPull reads sync.branch from SQLite
- Verify daemon reads sync.branch from SQLite config source
- Show config.yaml sync-branch is ignored (YAML-only key)
- Add skipped tracer bullet test for issue #831
Coverage: daemon autoPull config resolution
* fix(daemon): Use syncbranch.IsConfigured() for autoPull
Problem:
- Daemon's periodic sync never activated when sync-branch configured in config.yaml
- autoPull check only read sync.branch from SQLite, missing YAML-only configuration
Solution:
- Replace SQLite-only check with syncbranch.IsConfigured()
- Update test to validate correct detection of YAML-configured sync branch
Impact:
- Periodic sync now activates correctly when sync-branch is configured in config.yaml
- Fixes daemon hanging issue for users with YAML-only configuration
When .beads/redirect exists, bd sync was using GetRepoRoot(cwd) to find
the git worktree location. This failed because the worktree should be
in the same repo as the beads directory, not the current working directory.
Fix: Use getRepoRootFromPath(beadsDir) to derive the repo root from
the actual beads location after following the redirect.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Adds closed_by_session tracking for entity CV building per Gas Town
decision 009-session-events-architecture.md.
Changes:
- Add ClosedBySession field to Issue struct
- Add closed_by_session column to issues table (migration 034)
- Add --session flag to bd close command
- Support CLAUDE_SESSION_ID env var as fallback
- Add --session flag to bd update for status=closed
- Display closed_by_session in bd show output
- Update Storage interface to include session parameter in CloseIssue
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Executed-By: beads/crew/dave
Rig: beads
Role: crew
For molecules with >100 steps, shows summary instead of full step list:
- Counts children first using efficient GetMoleculeProgress query
- Shows progress summary with pointer to bd mol progress
- Add --limit flag to cap output (e.g., --limit 50)
- Add --range flag for specific ranges (e.g., --range 1-50)
Also closes epic bd-5nu1 (mega-molecule progress support).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Replace findInProgressMolecules (which loads full subgraphs) with
findInProgressMoleculeIDs (which only returns IDs). This ensures
auto-discovery is efficient for mega-molecules.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Root cause: When a bd daemon crashes, its daemon.lock file remains with
the old PID. If that PID gets reused by an unrelated process, the code
would wait 5 seconds for a socket that will never appear.
Fix: Use flock-based check as authoritative source for daemon liveness.
The OS releases flocks when a process dies, so this is immune to PID reuse.
Changes:
- handleExistingSocket: Check daemon flock before waiting for socket
- acquireStartLock: Verify daemon lock is held before waiting
- handleStaleLock: Use flock check to detect stale startlocks
- lockfile/process_*.go: Add pid <= 0 check to prevent false positives
(PID 0 signals process group on Unix, not a specific process)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Warn if source issue is already closed
- Copy all issue fields (SourceRepo, Ephemeral, MolType, RoleType, Rig)
- Log warnings on label copy failures instead of silently ignoring
- Schedule auto-flush when source is local store
- Remove stale comment
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Users with conflicting git hooks (e.g., hooks that read the staging
area like GGA) can now set BEADS_NO_AUTO_STAGE=1 to disable auto-staging
and get check-and-block behavior instead.
Default behavior unchanged - auto-staging still works for most users.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Adds a new `bd mol progress` command that shows molecule progress using
indexed queries instead of loading all steps into memory. This makes it
suitable for mega-molecules with millions of steps.
Features:
- Efficient SQL-based counting via idx_dependencies_depends_on_type index
- Progress display: completed / total (percentage)
- Current step identification
- Rate calculation from closure timestamps
- ETA estimation
- JSON output support
New storage interface method: GetMoleculeProgress(ctx, moleculeID)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implements `bd refile <source-id> <target-rig>` which:
- Gets source issue via routing
- Creates new issue in target rig with same content
- Closes source with "Refiled to <new-id>"
- Appends "(Refiled from <old-id>)" to description
Target rig accepts rig names (beads), prefixes (bd-), or prefix shorthand (bd).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
