Add author handle to display name and missing language tag for the
fancypantalons/nvim-beads entry to distinguish it from joeblubaugh/nvim-beads.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Documents that the standalone beads-merge binary is deprecated in favor of
bd merge, and adds Jujutsu (jj) integration instructions for resolving
beads merge conflicts.
PR: https://github.com/steveyegge/beads/pull/2318
Author: maphew
Two issues:
1. Epic child creation commands were missing the --parent flag,
making it appear that child tasks are automatically associated
with the most recently created epic. The --parent flag is required.
2. The dependency visibility example (bd list showing "blocked by"
annotations) was placed immediately after issue creation, before
any dependencies were added. Moved it after the "Add Dependencies"
section where it makes sense.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add Custom Merge Driver section documenting:
- Deprecation of standalone beads-merge binary (previously at neongreen/mono)
in favor of built-in bd merge command
- Updated Jujutsu (jj) integration config using bd as the merge program
Fixes#906
Amp-Thread-ID: https://ampcode.com/threads/T-019cb569-6bbd-73fd-85a8-0d5dcb78b711
Co-authored-by: Amp <amp@ampcode.com>
* refactor(testutil): consolidate Dolt readiness into enum
Expand from 3-state to 5-state doltReadiness enum:
- doltNoDocker (renamed from doltNoDaemon for clarity)
- doltNoImage (no image at all)
- doltWrongVersion (repo cached but wrong tag)
- doltSkipped (explicit BEADS_TEST_SKIP=dolt opt-out)
- doltReady
Add isDoltRepoImageCached() to distinguish "no image" from "wrong
version". Add hasTestSkip() for comma-separated BEADS_TEST_SKIP
parsing. Silent skip for explicit opt-out avoids ~1s docker info
overhead.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* docs: clarify scope of ddfc5e94ddfc5e94 title says "consolidate" but actually expands the design:
Code changes (internal/testutil/testdoltserver.go):
- Expand doltReadiness enum from 3 to 5 states
- Rename doltNoDaemon → doltNoDocker (clarity)
- Add doltWrongVersion state (image repo cached, wrong tag)
- Add doltSkipped state (explicit opt-out)
- Introduce BEADS_TEST_SKIP env var (comma-separated service list)
replacing the owner's implicit Docker-only detection with an
explicit opt-out mechanism that skips the ~1s docker info cost
- Add isDoltRepoImageCached() helper (docker images <repo> -q)
- Add hasTestSkip(service) helper (parses BEADS_TEST_SKIP)
- Extract doltDockerRepo package var to deduplicate strings.Cut
- Silent skip for doltSkipped (no WARN in StartTestDoltServer)
Documentation (docs/TESTING.md):
- Add 5-state readiness table
- Document BEADS_TEST_SKIP usage and evaluation order
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Embedded Dolt is being reintroduced for standalone/solo Beads users
(Dolt team delivering improved embedded support). Update docs and code
comments that asserted embedded was permanently removed. Gas Town
continues to use server mode exclusively.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Dolt is now the default (and only) storage backend for bd init. Add a
--backend flag that accepts "dolt" (default) or "sqlite" (prints a
deprecation notice with migration instructions and exits). Unknown
backends produce a clear error.
Update QUICKSTART.md and DOLT-BACKEND.md to reflect that SQLite is no
longer supported and Dolt is the sole backend.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds --stdin boolean flag to bd create and bd update that reads the
description from stdin. This is a natural desire path -- agents frequently
try --stdin before discovering --body-file -. Uses cobra
MarkFlagsMutuallyExclusive to prevent conflicts with other description flags.
Adds `bd show --current` which resolves the currently active issue using
priority: in-progress assigned to actor > hooked > last touched. Exits
non-zero when no current issue is found, fixing the exit code problem
where unrecognized flags silently exited 0 with help text.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- #2250: setupStealthMode skips git-exclude when not in a git repo
- #2239: remove dead Monitor WebUI reference from COMMUNITY_TOOLS.md
- #2183: add "holder" to isAllowedUpdateField so merge-slot acquire/release works
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Executed-By: beads/crew/emma
Rig: beads
Role: crew
bd dolt remote add/remove/list now operate on both the SQL server and
filesystem CLI config, ensuring dolt push/pull via CLI can always find
the remote.
- remote add: writes both surfaces, prompts before overwriting
- remote list: shows both surfaces, flags discrepancies
- remote remove: removes both, refuses on URL conflict (--force to override)
- dolt show: includes remotes section
- dolt test: tests remote connectivity (SSH/HTTP)
- doctor: CheckRemoteConsistency in Data & Config, fixable for one-side-only
(conflicts skipped even with --yes)
- docs: update DOLT-BACKEND.md with dual-surface behavior and refs/dolt/data note
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Executed-By: beads/crew/emma
Rig: beads
Role: crew
CALL DOLT_PUSH/DOLT_PULL through the SQL server times out for SSH
remotes — the MySQL connection drops before the SSH transfer completes.
Detect SSH remote URLs (git+ssh://, ssh://, git@) and shell out to
`dolt push`/`dolt pull` from the database directory instead. Hosted
Dolt, DoltHub, S3, GCS, and file remotes continue using SQL as before.
Also updates DOLT-BACKEND.md to document `bd dolt remote add` (added
in 573bdd1b) instead of the stale `cd .beads/dolt && dolt remote add`
pattern. Remotes added via CLI are not visible to the running SQL
server, causing the "remote 'origin' not found" error in #2118.
Tested with git+ssh:// remotes to GitHub on two standalone repos.
Other remote types (Hosted Dolt, S3, etc.) are untested by this change.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Executed-By: beads/crew/emma
Rig: beads
Role: crew
- Add metric naming convention table (SDK dot-notation → Prometheus underscores)
- Move doltserver.* events to roadmap — zero OTel imports in doltserver package
- Clarify hook.exec as span-only (no metric counters or histograms)
- Flag bd.db.lock_wait_ms as registered but never recorded
- Add circuit breaker metrics (bd.db.circuit_trips/rejected)
- Add source reference appendix: every span, metric, and attribute
backed by file:line to prevent documentation drift
- Correct Init()/Shutdown() signatures to match actual telemetry.go
- Fix SQL span section to reflect actual dolt.query/exec/query_row spans
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
DOLT.md: fix data location paths to ~/gt/.dolt-data/
dolt-concurrency.md: status from draft to implemented
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- findGitRoot: use exec.CommandContext with 5s timeout to prevent hangs
- store.go nolint: clarify that safety comes from escaping, not validator
rejecting wildcards (underscores ARE allowed in DB names)
- TROUBLESHOOTING.md: replace ambiguous "this version" with changelog refs
Authored by: Aaron Lippold<lippold@gmail.com>
bd unconditionally ran CREATE DATABASE IF NOT EXISTS on every connection.
When connecting to the wrong server (auto-started from .beads/dolt/ or
Homebrew default), this created an empty shadow database hiding real data.
Changes:
- Add CreateIfMissing field to Config — only bd init sets it true
- SHOW DATABASES LIKE check before CREATE (with underscore escaping)
- Network error handling for SHOW DATABASES failure
- Suppress auto-start when metadata.json has explicit dolt_server_port
- Fix backup git push across .beads/redirect (findGitRoot)
- Fix macOS CanonicalizePath case resolution (walk dirs, not realpath)
- Add CreateIfMissing to all migration and create paths
- 10 regression tests covering guard, auto-start, ReadOnly, underscore
- CHANGELOG and TROUBLESHOOTING docs
Authored by: Aaron Lippold<lippold@gmail.com>
Add maybeAutoPush() in PersistentPostRun (after auto-commit and auto-backup)
that pushes to the Dolt remote when an origin is configured. Uses the same
pattern as maybeAutoBackup: debounce interval, change detection via commit
hash comparison, and non-fatal warnings on failure.
Config keys: dolt.auto-push (bool, auto-enabled when origin exists),
dolt.auto-push-interval (duration, default 5m).
This completes the Dolt replication story: users add a remote once and
data flows automatically.
Closes: bd-5x9
Executed-By: beads/polecats/quartz
Rig: beads
Role: polecats
Document the full dependency and gate systems which had no dedicated
guide. Prompted by community discussion #2164 asking how to keep
dependent work in sync when beads updates aren't tied to code merges.
- New docs/DEPENDENCIES.md: dependency types, ready work, visualization,
cross-repo deps, gate types (gh:pr, gh:run, timer, bead, human),
gate automation, and workflow recipes
- Add Gates section to CLI_REFERENCE.md (bd gate commands were missing)
- Cross-reference from QUICKSTART.md and DOLT.md
Adds `bd backup` command that exports all database tables to sorted JSONL
files in .beads/backup/ for off-machine recovery. Events are exported
incrementally using a high-water mark. Each table is written atomically
via temp file + rename.
Auto-backup runs in PersistentPostRun when backup.enabled is true,
throttled by backup.interval (default 15m). Optional git-push mode
auto-commits and pushes backup files.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Executed-By: beads/crew/emma
Rig: beads
Role: crew
Updates AGENT_INSTRUCTIONS.md and docs/CLAUDE.md to recommend make install over go build -o bd, preventing stale binary issues.
PR: https://github.com/steveyegge/beads/pull/2109
Author: l0g1x (Krystian Gebis)
Agent instructions and CLAUDE.md told agents to run
`go build -o bd ./cmd/bd` followed by `./bd`, creating stale local
binaries in every worktree. Change to `make install` which places the
binary in ~/.local/bin (the canonical location) and use `bd` from PATH
instead of `./bd`.
Standardize start-work semantics to `bd update <id> --claim` across docs,
examples, templates, and website static LLM artifacts. Replaces
`--status in_progress` with `--claim` where the meaning is start-work.
Also fixes stale comment in llms.txt ("Update status" → "Start work").
Co-Authored-By: Valeriy Pavlovich <iqdoctor@users.noreply.github.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add a new MCP `claim` tool as the atomic start-work operation (CLI equivalent: `bd update <id> --claim`). Keep `update` for semantic updates. Update MCP-facing docs/examples to use `claim` for start-work. Add focused tests for client/tool/server claim paths.
Co-Authored-By: Valeriy Pavlovich <iqdoctor@users.noreply.github.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* feat: support issue_id_mode=counter for sequential IDs (GH#2002)
When `issue_id_mode=counter` is set via `bd config set issue_id_mode counter`,
`bd create` now assigns monotonically increasing integer IDs instead of
hash-based IDs. The counter is stored in a new `issue_counter` table (one row
per prefix) and incremented atomically within the same transaction as the issue
insert.
- Add `issue_counter` table to schema (version bumped to 5)
- Add migration 006 to create `issue_counter` table on existing databases
- Add `GetNextIssueCounter()` to queries.go (standalone transaction, for
callers outside CreateIssue)
- Modify `generateIssueID()` to check `issue_id_mode` config and use counter
path when set to "counter"; hash mode remains the default when unset
- Explicit `--id` flag continues to take precedence over counter mode
- Tests: TestGetNextIssueCounter_Sequential, TestGetNextIssueCounter_MultiplePrefixes,
TestCreateIssue_CounterMode, TestCreateIssue_ExplicitIDOverridesCounter,
TestCreateIssue_HashModeDefault, TestMigrateIssueCounterTable
* docs: document issue_id_mode=counter feature (GH#2002)
Add documentation for the sequential counter ID mode across all relevant docs:
- docs/CONFIG.md: new issue_id_mode namespace entry and full example section
with tradeoff table, migration guidance, and per-prefix counter isolation
- .beads/BD_GUIDE.md: counter mode section with when-to-use, migration
considerations, and explicit --id override behavior
- docs/ADAPTIVE_IDS.md: alternative counter mode section with cross-reference
to CONFIG.md
- website/docs/reference/configuration.md: issue_id_mode under ID Generation
with comparison table
* fix: seed counter from existing issues when enabling counter mode (GH#2002)
When issue_id_mode=counter is enabled on a repo that already has manually-created
sequential IDs (e.g., plug-1 through plug-50), the counter used to start at 1,
causing immediate collisions.
Add seedCounterFromExistingIssuesTx() which scans existing issue IDs, finds the
highest numeric suffix for the given prefix, and seeds the issue_counter table
from there. The function is idempotent (skips if a counter row already exists)
and only counts purely-numeric suffixes (ignoring hash-based IDs like test-a3f2).
The seeding is called at first counter use: in generateIssueID() (issues.go) and
GetNextIssueCounter() (queries.go) when sql.ErrNoRows is returned for the prefix.
Add four tests covering: seeding from existing numeric IDs, mixed hash+numeric
IDs, fresh repos (counter starts at 1), and already-seeded counters (no regress).
* feat: add bd config schema and describe commands for agent discoverability (GH#2002)
Adds machine-readable config schema so agents can programmatically discover
all available configuration keys without reading source code.
- internal/config/schema.go: new ConfigKeyDef struct and Schema slice with
43 entries covering core, sync, routing, jira, linear, gitlab, team, mail,
and YAML-only keys; includes type, default, valid values, description, and
storage location per key
- cmd/bd/config.go: adds 'bd config schema' (table + --json) and
'bd config describe <key>' (single-key detail + --json) subcommands
Usage:
bd config schema
bd config schema --json | jq '.[] | select(.key == "issue_id_mode")'
bd config describe issue_id_mode
bd config describe jira.url --json
* refactor: remove overengineered config schema (keep docs only)
Remove the configSchemaCmd and configDescribeCmd subcommands added in
the previous commit, along with internal/config/schema.go (390 lines of
config key definitions). These are out of scope for GH#2002.