hrt-dark-mode-opt-in
Opt-in gate for tempa-spec's dark mode (bounded autonomous decision-making across proposal-through-align, optionally through apply). Never offered proactively - only activates when the user asks for it unprompted, in their own words. Shows a mandatory risk warning before any confirmation, then asks which variant (planning-only or +implementation) and creates dark.md. Use once, at or after the proposal interview, only on explicit user request.
How do I install this agent skill?
npx skills add https://github.com/hash-id/openspec-schema --skill hrt-dark-mode-opt-inIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill establishes an autonomous operating mode that reduces manual human-in-the-loop oversight for the agent's decisions. While it includes mandatory risk disclosures, explicit opt-in requirements, and maintains manual stops for security-critical actions, the increase in autonomy expands the impact of potentially incorrect agent decisions.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
What does this agent skill do?
hrt-dark-mode-opt-in
Lets the agent resolve findings that would normally escalate to the user - across specs, design, align, optionally apply - via hrt-dark-mode-decision-gate, logging every decision to dark.md. WALKTHROUGH stays mandatory in every variant.
Never proactive
MUST NOT offer, suggest, or mention dark mode - at any phase, in any form. This skill MUST NOT run unless the user has already asked for dark mode unprompted, in their own words. If they never ask, this skill never runs and nothing changes.
Steps
-
User asked for dark mode. Show the warning below, in full, unedited, before anything else.
-
Ask which variant:
- Planning-only. Covers
proposalthroughalign. WALKTHROUGH stays a fully manual closing review.applyis out of scope for this run. - +implementation. Same, then continues automatically into
applyin the same session after a clean WALKTHROUGH, usinghrt-dark-mode-decision-gateforapply's review findings too. A second closing review happens afterapply, showing the code diff alongsidedark.md.
State that +implementation carries every planning-only risk plus the risk that a bad decision reaches shipped code, not just a document.
- Planning-only. Covers
-
MUST get an explicit answer naming the variant before proceeding. MUST NOT treat silence, or an answer to a different question, as that answer.
-
Create
dark.mdfrom thedark.mdtemplate: chosen variant + timestamp in the header, per-stage sections empty. Include the Apply section only if +implementation. -
dark.md's existence is the only signal dark mode is active. No separate flag.
The warning (show verbatim, in full, before step 2)
DARK MODE is a PROTOTYPE and DANGEROUS. Do not use it for sensitive or critical work. The agent will make big decisions by itself. Normally, the agent stops and asks you before a big decision.
In dark mode, it does not stop. It decides alone, and it can decide wrong. It still stops and asks you about security decisions, and about steps that cannot be undone easily (for example a data migration, deleting data, or changing a published API). A wrong decision can damage your codebase. In the worst case, your work could be lost.
If something goes wrong, YOU ARE RESPONSIBLE for the result.
If step 3's confirmation was given without this warning shown first, in full, MUST NOT treat it as valid - show the warning and ask again.
Follows the plain-language-writing skill's COMMS rules.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/hash-id/openspec-schema/hrt-dark-mode-opt-in">View hrt-dark-mode-opt-in on skillZs</a>