setup
Skill93 starsChanged 57 days agoMITView on GitHub
Configure a project's Dely pins — per-phase harness, model and effort for implement and review, written to .dely/pins.json (team) or .dely/local/pins.json (only this person), discovered from the live harness surface. Use at the start of a Control Session, when the project has no pins, or when pins need rewriting from the installed harnesses. Not for installing plugins, trusting hooks, or delivering a change; that is delivery.
191 other projects use a version of this.
Adopted by
Most recently taken up by
- hydra-db/hydradb-claude-codetheir own version, first seen Nov 8, 2026
- WALKERKILLER/Look-Tongji-Notestheir own version, first seen Nov 8, 2026
- steerlabs/opensteertheir own version, first seen Nov 8, 2026
- ansvisor/ansvisortheir own version, first seen Nov 8, 2026
- dsifry/metaswarmtheir own version, first seen Nov 8, 2026
- anombyte93/prd-taskmastertheir own version, first seen Nov 8, 2026
Used with
Projects with it also use add-voice-transcription, add-discord and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
juancruzmunozalbeloSkilladd-voice-transcription
Add voice message transcription to NanoClaw using OpenAI's Whisper API. Automatically transcribes WhatsApp voice notes so the agent can read and respond to them.
Also in 5 projects that use this
MicrosoftSkilladd-discord
Add Discord bot channel integration via Chat SDK.
Also in 11 projects that use this
MicrosoftSkilladd-slack
Add Slack channel integration via Chat SDK.
Also in 11 projects that use this
juancruzmunozalbeloSkilladd-telegram-swarm
Add Agent Swarm (Teams) support to Telegram. Each subagent gets its own bot identity in the group. Requires Telegram channel to be set up first (use /add-telegram). Triggers on "agent swarm", "agent teams telegram", "telegram swarm", "bot pool".
Also in 7 projects that use this
- +8 more from microsoft
Other versions191 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 191
- anthropics/knowledge-work-plugins★ 35ksame name
- garrytan/gbrain★ 32ksame name
- gavrielc/nanoclaw★ 31ksame name
- nanocoai/nanoclaw★ 31ksame name
- qwibitai/nanoclaw★ 31ksame name
- Q00/ouroboros★ 6.3ksame name
- pytorch/executorch★ 5.2ksame name
- agenticnotetaking/arscontexta★ 3.5ksame name
- aws/agent-toolkit-for-aws★ 3.4ksame name
- spotify/portal-ai-plugins★ 2.8ksame name
181 more not listed
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.
The file
Show its rules and text
We split this skill into the rules it gives an agent
14 rules. 0 are common, 3 are distinctive, and 11 are very specific to this project.
0
Common
Most of the 170 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
11
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Write the project's Dely pins into one JSON file, and one routing sentence into AGENTS.md.
- Very specific: line 41
Quick: Use the current harness for both implement and review.
- Distinctive: line 53
Ask which path, and whether the choice is for the team or only for this person.
- Very specific: line 70
Use the harness id, not its display name.
- Very specific: line 80
If AGENTS.md contains an old <!-- dely:begin --> block, the helper no longer reads it.
- Very specific: line 102
When discovery.answeredLocally is true, the probe is answered locally: do not treat it as a dispatch.
- Very specific: line 105
When discovery.omitVisibility is set, slugs whose visibility equals that value are not offered.
- Very specific: line 109
When discovery.modelField is set, each model's value is that field rather than slug.
- Very specific: line 112
When effortFlag is false and discovery.effort and discovery.effortFrom are both absent, write the literal default for Effort.
- Distinctive: line 145
Stop and report to the human, unchanged, when an existing pins file is not valid JSON in the format above.
- Very specific: line 150
Read the current harness's instructionsFile.
- Very specific: line 168
After the pins are written, for each pinned harness whose trust is dialog, open it once for the human with orca terminal create --worktree …
- Very specific: line 172
If permissionDefault is none, the command is the binary alone.
- Distinctive: line 186
No plugin or skill install.
“Common” is measured: kept by at least 6 in 10 of the 170 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
Setup
Write the project's Dely pins into one JSON file, and one routing sentence
into AGENTS.md. Discover models and effort from the installed harnesses. Do
not store a catalogue. Do not install, trust, or enumerate anything.
Harness facts live in ../../harnesses.json relative to this skill
(repository root harnesses.json). Read that file. Do not copy a list from
this package, from memory, or from docs/.
Two files
.dely/pins.json team pins, tracked
.dely/local/pins.json personal pins, never tracked
.dely/local/.gitignore contains *, so nothing under .dely/local is tracked
.dely/local/runs/<run>/ prompts and handoffs of one Orca Run, written by deliveryEach file is an object whose only keys are implement and review. Each
value has exactly the string keys harness (an id from harnesses.json),
model and effort. A file may name one phase or both. Per phase, the
personal file wins over the team file, and the team file over Control's own
harness with default for Model and Effort. In a linked worktree, the
personal file is looked for in that worktree first and then in the main
checkout; only one personal file is read, never both. dely pins --repo <path>
prints what resolves and from where.
Pins configure implement and review — nothing else. There is no control
row, because the current interactive session already exists and is never
dispatched; there is no release row, because release has no LLM worker; there
is no Plan Mode or design-skill field.
Two paths
Quick. Use the current harness for both implement and review. Use
that harness's defaults for model and effort, written as the literal
default.
Customize. For each of implement and review, offer the discovered
harnesses, models and effort levels and write what the human chooses. Where a
harness exposes no way to pin a model or an effort, write the literal default
for that cell and point to Orca's agent default arguments. An entry with
effortInModel or modelPin: spec-unchecked can be pinned — offer its
discovered models and effort levels as for a flag harness, and write
Effort default for a model whose effort field is null.
Ask which path, and whether the choice is for the team or only for this person. Do not start writing until both are answered.
What to write
For a team choice, write .dely/pins.json. For a personal choice, write
.dely/local/pins.json, and write .dely/local/.gitignore containing the
single line * first, so the personal file never shows in git status. Never
edit the repository's root .gitignore. Create the directories as needed.
{
"implement": { "harness": "claude", "model": "default", "effort": "default" },
"review": { "harness": "codex", "model": "default", "effort": "default" }
}Use the harness id, not its display name. If the target file exists, show
what it holds and write only on a yes. A personal file shadows the team file
per phase; say so when writing one.
Then make sure AGENTS.md carries the routing line. If it has no
dely:delivery, append:
Bounded or Architectural work invokes
dely:delivery; Spike starts no delivery run.
If AGENTS.md contains an old <!-- dely:begin --> block, the helper no
longer reads it. Offer to replace the block, from its begin marker through its
end marker, with the routing line, and carry its table over to
.dely/pins.json as the proposed team pins. Change AGENTS.md only on a yes.
Prose outside the block is never touched.
Discovery
Offer only entries whose status is supported. A deferred entry is
omitted, not an error. When an offered entry has limits, print it next
to that harness.
For each supported entry whose binary is installed, run that entry's
discovery. discovery is null when the harness has no listing command;
then write the literal default for Model and Effort and point the human to
Orca's agent default arguments to set the model. When discovery is an
object, run discovery.models and, when discovery.effort is set, that
command too.
They are the catalogue. Do not invent a catalogue, do not prompt a model to
learn one, and do not treat a /model slash command as discovery.
When discovery.answeredLocally is true, the probe is answered locally: do
not treat it as a dispatch.
When discovery.omitVisibility is set, slugs whose visibility equals that
value are not offered. When discovery.effortFrom is set, effort levels are
that field on each slug, not one vocabulary per harness. When
discovery.modelSlugBefore is set, offer the slug before that separator.
When discovery.modelField is set, each model's value is that field rather
than slug.
When effortFlag is false and discovery.effort and discovery.effortFrom
are both absent, write the literal default for Effort. Do not invent an
effort vocabulary, do not strip effort suffixes from slugs, and do not
synthesize parameterized [effort=…] forms. Omit discovery that is
unavailable or unusable rather than guessing.
A harness that is not installed is omitted from the offer, not an error.
Pinning
The literal default in Model or Effort means the harness default is wanted:
omit that flag. That is not the same as an unset phase: defaults are a
deployment preference, not a reproducible pin, and the execution envelope
records the configured value and, where the harness exposes it, the actual
observed model and effort.
Where no file names a phase, delivery runs it on Control's own harness with
default for Model and Effort. Setup is a convenience over that fallback, not
a precondition for it. The run stops only when Control's harness is unknown.
When a choice cannot be offered
Where a choice cannot be offered, do not make it. Take the conservative action — which may be writing a conservative value, and may be doing nothing — and report the choice that was not offered, naming what was available and how to set it; do not write that report into a pins file.
For the instructions-file import: if the current harness's
instructionsFile.needsImport is true, the import is absent, and setup
cannot ask, write nothing and report the offer that was not made.
Refusals
Stop and report to the human, unchanged, when an existing pins file is not valid JSON in the format above. Do not repair it or guess what it meant.
Instructions file
Read the current harness's instructionsFile. The persistent instruction
reaches a harness natively when readsAgentsMd is true. It reaches a
harness that does not read AGENTS.md only where the project has
instructionsFile.file containing instructionsFile.importLine.
Where needsImport is true and that import is absent, offer to create a
one-line file at instructionsFile.file containing
instructionsFile.importLine. The human accepts or declines. Never write it
unasked.
This carries no configuration: it points at AGENTS.md, it does not copy it.
The offer follows needsImport. Files listed in inert are not imported.
Files listed in alsoApplies are applied as rules by that harness; that
does not by itself trigger the write offer.
Trust
After the pins are written, for each pinned harness whose trust is
dialog, open it once for the human with
orca terminal create --worktree path:<repo> --command "<binary> <permissionDefault>",
taking binary and permissionDefault from that entry in harnesses.json.
If permissionDefault is none, the command is the binary alone. The human
answers that harness's own dialog; setup never answers it and never writes a
harness store. The human closes the terminal when done. orca-preflight and
none need no step.
Preflight
Open a Run first as orca skills get orchestration describes. Then run
../delivery/scripts/dely preflight --repo <path> --run <runId> relative
to this skill. Any PREFLIGHT … FAIL (exit 1): do not dispatch to any pin;
relay the printed reason to the human.
What setup will not do
No plugin or skill install. No hook trust. Setup may open a pinned harness in an Orca terminal so the human can answer that harness's own trust dialog; it still never answers the dialog and never writes a harness store. No custom agent creation or modification. No coordinator installation or field. No control or release row. No enumeration or invocation of project-owned workflow plugins. No model catalogue.
Print verified install guidance only when the human explicitly asks for it.
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.

