inception
Skill1 starsChanged 10 days agoMITView on GitHub
Produces a durable, file-level implementation plan by loosely composing Balakit specialists. Stops before code except --autopilot, which then runs execute. Default interactive; --circuit skips mid-plan waits and silent-stops at the plan. Apply when the user invokes inception or /inception, or wants a long-running detailed plan. If host Plan Mode is on, write into that native plan. If not, write .balakit/plans/<slug>.md. Use dissect to audit an existing plan. Use deep-deliberation to pick among approaches. Use execute to implement.
6 other projects use a version of this.
Adopted by
Most recently taken up by
- inkatze/planwrighttheir own version, first seen Nov 1, 2026
- steepy-it/steepy-apextheir own version, first seen Nov 1, 2026
- steepy-it/steepy-formulatheir own version, first seen Oct 25, 2026
- XyndoX/good-bad-uglytheir own version, first seen Oct 13, 2026
- arthur422tp/Inceptiontheir own version, first seen Aug 29, 2026
- mauriciovieira/groundworktheir own version, first seen Jul 10, 2026
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- steepy-it/steepy-apex★ 14same name
- steepy-it/steepy-formula★ 5same name
- inkatze/planwright★ 4same name
- arthur422tp/Inception★ 3same name
- XyndoX/good-bad-ugly★ 1same name
- mauriciovieira/groundwork★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository107 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .cursor/rules/base.mdcCursor rule
- .cursor/rules/changelog.mdcCursor rule
- .cursor/rules/comments.mdcCursor rule
- .cursor/rules/release.mdcCursor rule
- .cursor/rules/seo-ai-search.mdcCursor rule
- .cursor/rules/testing.mdcCursor rule
- agent-readySkill
- authoring-skills-and-rulesSkill
- blinded-evalSkill
- cloakbrowser-fallbackSkill
- deep-deliberationSkill
- design-rationaleSkill
- dissectSkill
- documentation-writerSkill
- everything-seoSkill
- executeSkill
- generating-app-verifySkill
- kit-workflowsSkill
- marketing-psychologySkill
- media-genSkill
- nlm-skillSkill
- opinionSkill
- proving-change-safetySkill
- refreshing-app-verifySkill
- release-deploySkill
- seo-auditSkill
- startup-marketing-brainSkill
- subsystem-walkthroughSkill
- trlSkill
- unmatched-workflowSkill
- agent-readySkill
- authoring-skills-and-rulesSkill
- blinded-evalSkill
- cloakbrowser-fallbackSkill
- deep-deliberationSkill
- design-rationaleSkill
- dissectSkill
- documentation-writerSkill
- everything-seoSkill
- executeSkill
- generating-app-verifySkill
- inceptionSkill
- kit-workflowsSkill
- marketing-psychologySkill
- media-genSkill
- nlm-skillSkill
- opinionSkill
- proving-change-safetySkill
- refreshing-app-verifySkill
- release-deploySkill
- seo-auditSkill
- startup-marketing-brainSkill
- subsystem-walkthroughSkill
- trlSkill
- unmatched-workflowSkill
- agent-readySkill
- authoring-skills-and-rulesSkill
- blinded-evalSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
9 rules. 0 are common, 6 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
6
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 47
Default is interactive: ask when blocked or the choice is costly.
- Distinctive: line 69
Stop and ask even on autopilot.
- Distinctive: line 78
Do not auto-start this skill.
- Very specific: line 80
Do not implement product code.
- Very specific: line 81
Do not SwitchMode / enter Plan Mode unless they chose that this turn.
- Very specific: line 82
Do not name this skill plan and do not ship a second OpenCode primary plan.
- Distinctive: line 90
Start every pipeline response with this block and update it mechanically:
- Distinctive: line 102
If the state is missing after a context change, reconstruct it from the latest canvas path and ask the user to confirm before advancing.
- Distinctive: line 113
Write the plan onto the canvas.
“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Inception
Leading words: host canvas, one canvas, specialist catalog, depth bar, skip the wait not the work, circuit, autopilot.
Inception authors a plan. It does not implement — except --autopilot, which
writes the plan then runs execute in this chat.
User input
$ARGUMENTSParse a pace flag before treating the rest as the request:
--circuit, --autopilot, --auto (alias of --autopilot). Strip the flag
from the request text. If both circuit and autopilot appear, autopilot wins.
Pace modes (optional)
Default is interactive: ask when blocked or the choice is costly.
The user must opt in this invocation — a flag above, or an explicit phrase
(circuit, autopilot, take recommended, circuit the rest,
autopilot the rest). Do not infer from urgency. Mid-run, the same phrases
switch mode from the current checkpoint. stop circuit / stop autopilot
drops back to interactive.
| Mode | Mid-plan waits | At the plan |
interactive | Ask. Clarify. Wait. | Write plan. Stop. Recommend execute. |
--circuit | Auto-accept recommended. PACE_LOG. | Write plan. Stop. Recommend execute. No extra ask. |
--autopilot | Auto-accept. PACE_LOG. | Write plan. Run execute (auto-pick distinct families; execute's interactive default is Current-for-all). |
This skill shares --circuit / --autopilot grammar with dissect and
deep-deliberation. The final gate differs: circuit silent-stops. Only
autopilot crosses into execute.
Skip the wait, not the work. Specialists still run when they apply. Emit a
compact PACE_LOG of every auto-accepted choice.
Cannot proceed: blocked fork, no unique plan, canvas cannot be written. Stop and ask even on autopilot.
Plan Mode vs autopilot: host Plan Mode cannot implement. Do not
SwitchMode / ExitPlanMode to run execute. Degrade to circuit.
skip: plan-mode in PACE_LOG. Tell them to accept / Build / leave Plan Mode,
then /execute.
Operating contract
- Do not auto-start this skill. The user must ask for it.
- One canvas. Never a sibling plan file.
- Do not implement product code.
--autopilothands off toexecute. - Do not
SwitchMode/ enter Plan Mode unless they chose that this turn. - Do not name this skill
planand do not ship a second OpenCode primaryplan. - Never write
.balakit/installed.json. - Pull specialists from the catalog with
skip: <reason>. Not a waterfall. - Preserve host-required plan structure (Cursor YAML todos, Claude injected path). Depth lands as sections, not a replacement template.
Runtime state
Start every pipeline response with this block and update it mechanically:
INCEPTION_STATE
stage: frame | specialists | write | done
pace: interactive | circuit | autopilot
host: cursor | claude-code | opencode | copilot | codex | other
plan_mode: on | off
canvas: unset | <path or host-uri>
next_action: one action onlyIf the state is missing after a context change, reconstruct it from the latest canvas path and ask the user to confirm before advancing.
Pipeline
Copy these steps into the todo list verbatim. Skip with skip: <reason>.
- Parse the pace flag. Detect the host and whether Plan Mode is on. Read references/hosts.md.
- Resolve the canvas (below). Record
canvasinINCEPTION_STATE. - Frame: outcome, constraints, non-goals, named data shape if it is already knowable. Interactive: ask only when blocked or the choice is costly.
- Pull specialists from references/specialists.md. Each unused row stays in the list with
skip: <reason>. - Write the plan onto the canvas. Meet references/depth-bar.md. Host-native format when Plan Mode is on.
- Interactive /
--circuit: stop. Recommendexecute.--autopilot: ifplan_mode: on, degrade (skip: plan-mode); else runexecutewith auto-pick roster.
Canvas
Plan Mode on → the live host plan is the document of record. Empty/new:
fill via the host-native create path. Already has content: patch in place
(KEEP / PATCH / APPEND). Do not full-file Write a replacement. Do not
create .balakit/plans/ as a second copy.
Plan Mode off → do not yank them into Plan Mode.
Interactive, ask:
prompt: Where should this plan live?
options:
- .balakit/plans/<slug>.md in this repo (Recommended)
- Switch to this host's Plan Mode (native UI)
- I'll pick a path
- Say this in plain English
allow_multiple: falseSay this in plain English is a meta question: explain native vs repo file
with one example, then re-ask. Do not write yet.
- Recommended →
.balakit/plans/<slug>.md(create the directory). Slug is kebab-case from the request, filesystem-safe. - Native UI →
SwitchModeto plan (user consent) or tell them Shift+Tab / Tab //plan, then treat Plan Mode as on. - I'll pick a path → they name it. Refuse
.balakit/installed.json,skills/, and a second file when a host plan is already live.
--circuit / --autopilot: skip the ask. Default .balakit/plans/<slug>.md.
Mental --against points at the canvas path. Do not copy the plan into Mental.
Reply contract
Playbook: inception
pace: interactive | circuit | autopilot
canvas: <path or host-uri>
specialists: <pulled>, skipped: <name (reason)>, …Then: the plan lives at canvas. Interactive and circuit recommend execute
next (not exclusive: kit-workflows / unmatched-workflow may still consume
it). Autopilot continues into execute unless skip: plan-mode.
Interactive final footer, last block, verbatim. Omit under --circuit /
--autopilot. Do not paraphrase.
⚡ **Skip the waits next time**
This run was **interactive** — questions when blocked. Same skill, two other paces:
- `--circuit` — skip mid-plan waits, write the plan, stop. Recommend execute. No extra ask.
- `--autopilot` — skip waits, write the plan, run execute (auto-pick). Degrades to circuit if Plan Mode is on.
`/inception --circuit <request>` · `/inception --autopilot <request>`References
| File | Use it for |
| references/hosts.md | Host detection, Plan Mode signals, native write path, forbidden fights |
| references/depth-bar.md | What “extremely detailed” means; required plan sections |
| references/specialists.md | Sibling skill catalog with apply-when and skip reasons |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

