Skip to content

Releases: protonspy/csdd

v0.2.18

13 Jun 16:43
82cdea9

Choose a tag to compare

feat(skills): add spec-brainstorm — a question-driven on-ramp to csdd…

v0.2.17

12 Jun 22:30
692f513

Choose a tag to compare

feat(spec): per-spec development_flow (unit | tdd | tdd-e2e) (#32)

* feat(update): preserve model and effort fields during updates for managed agents and skills

* feat(spec): add per-spec development_flow (unit|tdd|tdd-e2e)

Let a spec choose its development discipline at `spec init` so simple features
skip disproportionate test-first ceremony while critical ones keep full rigor.

- spec.json gains `development_flow` (omitempty): unit (tests after code),
  tdd (test-first RED→GREEN, default), tdd-e2e (TDD + end-to-end).
- `spec init --flow` across CLI, the MCP `csdd_spec_init` tool, and the TUI
  wizard, all via one exported `SpecInit` op (single source of truth).
- Workspace default resolved from steering frontmatter `default_development_flow`
  (absent ⇒ tdd); the steering validator flags an invalid value (exit 2).
- Backward compatible: a legacy spec.json with no field reads as tdd, and
  omitempty keeps untouched files byte-identical.
- Flow-aware guidance prose in the shipped baseline (tasks-generation,
  definition-of-done, tdd-cycle, implementer, tasks.md), plus a
  one-test-per-observable-behavior rule to curb test bloat.

Verification: gofmt clean · go vet ./... · go test -race ./... green
(455 pass) · tsc --noEmit (mcp-server) · spec validate passed.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

v0.2.16

09 Jun 15:13
69ba0bc

Choose a tag to compare

docs(baseline): make CLAUDE.md enforce the SDD flow instead of sugges…

v0.2.15

09 Jun 01:44
a30ef05

Choose a tag to compare

feat(web): order dashboard specs by created_at, newest first (#30)

* feat(update): preserve model and effort fields during updates for managed agents and skills

* feat(web): order dashboard specs by created_at, newest first

The web overview listed specs alphabetically by feature name. Order them
by spec.json created_at descending instead, so the most recently created
spec sits at the top of the panel.

created_at is an RFC3339 UTC string (always …Z, written once at `spec
init`), so a lexical compare is chronological. Equal created_at — and
unreadable specs, whose created_at is empty — fall back to feature name
ascending for a stable, total order. All three surfaces (API, TUI, CLI)
go through LoadOverview, so they stay consistent.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

v0.2.14

09 Jun 00:47
ffb6632

Choose a tag to compare

feat(baseline): clear context and branch-first at spec boundaries (#29)

Fold two spec-boundary discipline steps into the shipped baseline templates:
- /clear at each feature/spec boundary — start a spec on a clean context and
  reset once its PR ships; the Stop hook nudges when a spec's tasks are all
  checked.
- Branch-first before implementation — git switch -c <type>/<feature> (kebab
  case), enforced by the PreToolUse hook blocking commit/push on the default
  branch.

Updates the CLAUDE.md workflow, the SDD guide loop, the wf-development agent,
and the test-before-stop / block-destructive hooks. No spec — direct workflow
adjustment requested by the user.

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

v0.2.13

08 Jun 23:02
6f1b151

Choose a tag to compare

fix(mcp): add authorization header for GitHub preset (#28)

Add headers support for remote MCP servers and configure the GitHub preset with Authorization: Bearer ${GITHUB_PAT}.

v0.2.12

08 Jun 22:13
1b1df2f

Choose a tag to compare

feat(update): preserve model and effort fields during updates (#27)

Preserve model and effort frontmatter overrides when csdd update refreshes managed agents and skills.

Treat these execution overrides as local metadata so they do not create conflicts or .old backups by themselves.

v0.2.11

08 Jun 20:58
9557ea1

Choose a tag to compare

fix(cli): anchor test-report testPaths to the spec folder determinist…

v0.2.10

08 Jun 20:39
7429f87

Choose a tag to compare

feat(mcp): add `mcp install`/`mcp presets` quick-install presets (#26)

Add a built-in MCP preset registry (context7, playwright, github) exposed as
`csdd mcp install <name>...` and `csdd mcp presets`, reachable from the CLI and
the TUI. A preset is a pre-filled `mcp add` — it funnels through the existing
MCPAdd write path, so duplicate detection, transport validation, and the
on-disk shape are identical to a manual add (no new validator rule). The github
preset stores only its URL; auth is the client's OAuth on first use (no token
written by csdd).

Wire MCP usage into the shipped baseline: Context7 (current library/API docs)
into dev-architecture, tdd-cycle, discovery-research, and the implementer agent;
a new frontend-e2e-qa Playwright skill plus a skippable e2e step in
verify-change; and a GitHub-MCP note in pr-review.

Spec: specs/mcp-presets (requirements/design/tasks approved).
Verification: make check green (gofmt + go vet + go test -race ./...);
429 tests pass; code-reviewer approved with no blockers.

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

v0.2.9

08 Jun 17:42
6cbfb64

Choose a tag to compare

feat(web): show agent/skill model and effort in the dashboard (#25)

Surfaces the model/effort frontmatter (added by #24) in the read-only web
dashboard. The session overview now reads `model`/`effort` for agents and
skills into the Artifact metadata, and ResourceView renders them in the header
alongside tools. Follow-up to agent-skill-model-effort.

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>