* Refactor visual brainstorming: browser displays, terminal commands
Replaces the blocking TaskOutput/wait-for-feedback.sh pattern with a
non-blocking model where the browser is an interactive display and the
terminal stays available for conversation.
Server changes:
- Write user click events to .events JSONL file (per-screen, cleared on
new screen push) so Claude reads them on its next turn
- Replace regex-based wrapInFrame with <!-- CONTENT --> placeholder
- Add --foreground flag to start-server.sh for sandbox environments
- Harden startup with nohup/disown and liveness check
UI changes:
- Remove feedback footer (textarea + Send button)
- Add selection indicator bar ("Option X selected — return to terminal")
- Narrow click handler to [data-choice] elements only
Skill changes:
- Rewrite visual-companion.md for non-blocking loop
- Fix visual companion being skipped on Codex (no browser tools needed)
- Make visual companion offer a standalone question (one question rule)
Deletes wait-for-feedback.sh entirely.
* Add visual companion offer to brainstorming checklist for UX topics
The visual companion was a disconnected section at the bottom of SKILL.md
that agents never reached because it wasn't in the mandatory checklist.
Now step 2 evaluates whether the topic involves visual/UX decisions and
offers the companion if so. Non-visual topics (APIs, data models, etc.)
skip the step entirely.
* Add multi-select support to visual companion
Containers with data-multiselect allow toggling multiple selections.
Without it, behavior is unchanged (single-select). Indicator bar shows
count when multiple items are selected.
Drop 'top-level sessions only' which may cause the model to treat
using-superpowers as a first-turn-only skill, breaking skill chaining
on follow-up turns.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Let the SUBAGENT-STOP gate in using-superpowers handle skill leakage
instead of per-template directives. This avoids blocking non-superpowers
skills that users may want subagents to use.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Addresses Jesse's review feedback on PR #450:
- Move inline routing table from using-superpowers to references/codex-tools.md,
leveraging Codex's native progressive disclosure for companion files
- Narrow SUBAGENT-STOP from "Do not invoke skills" to "Do not invoke
superpowers skills" so subagents can still use non-superpowers skills
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace verbose scope explanations with a direct override statement
that explicitly claims priority over prior guidance (i.e. the
using-superpowers 1% rule injected by Codex's skill discovery).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Codex subagents inherit filesystem access and can discover superpowers skills
via native discovery. Without guidance, they activate the 1% rule and invoke
full skill workflows instead of executing their assigned task.
- Add SUBAGENT-STOP gate check above the 1% rule in using-superpowers
- Add Codex dispatch routing table (spawn_agent/wait/close_agent)
- Add scope directives to all 4 subagent dispatch templates
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Testing showed native skill discovery works without the AGENTS.md
gatekeeper — using-superpowers bootstraps itself via SKILL.md
frontmatter. Install is now just clone + symlink, driven by
INSTALL.md. No Node.js dependency.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Use fileURLToPath() instead of manual URL pathname parsing to correctly
handle paths with spaces and special characters on all platforms.
Replace execSync rm/rmdir with fs.unlinkSync for stale symlink removal.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- INSTALL.md: add prerequisites, Windows note, verify step, clone
deletion in uninstall
- README.codex.md: fix Windows section (junctions not symlinks),
add description field guidance, consistent terminology
- install-codex.mjs: accurate link type labels (symlink vs junction)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replaces bootstrap CLI references with native discovery flow.
Install is now clone + run installer. Documents tool mappings,
personal skills path, and Windows junction fallback.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Two-step install: clone + run installer. Replaces the old manual
setup that required editing AGENTS.md by hand.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Creates symlink from ~/.agents/skills/superpowers to repo skills dir,
updates ~/.codex/AGENTS.md with gatekeeper block, removes old bootstrap
block if present. Windows junction fallback when symlinks are blocked.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The bootstrap CLI (superpowers-codex), Windows wrapper, and bootstrap
content file are no longer needed — Codex now has native skill discovery
that replaces this mechanism.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
PowerShell doesn't expand ~ when passed as an argument to node,
causing MODULE_NOT_FOUND errors. $HOME expands correctly in both
bash and PowerShell.
Fixes#285
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Windows doesn't respect shebangs, so directly invoking the extensionless
superpowers-codex script triggers an "Open with" dialog. Prefix all
invocations with `node` (harmless on Unix, required on Windows) and add
a .cmd wrapper for manual invocation on Windows.
Fixes#285, #243
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
