Skip to content
Skills Library
AI & agents · Automation

agent-native

/agent-native

Dispatch independent tasks to parallel workers or subagents without write collisions. Use when: running or planning agents in parallel, even two; check scopes before any launch.

Repository
agentops ↗
Licence
Apache-2.0
Last repository change
25.10.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
10
Scripts
1
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 25.10.2026, without references to retired models or APIs.

Licence

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

What to check before installing
  • 1 scripts in this package. Review them before installing, as Claude may execute them.

Scripts: scripts/fake_model_runner.py

What it does and when to use it

Dispatch independent tasks to parallel workers or subagents without write collisions. Use when: running or planning agents in parallel, even two; check scopes before any launch.

Original author description

Dispatch independent tasks to parallel workers or subagents without write collisions. Use when: running or planning agents in parallel, even two; check scopes before any launch.

How it works

  1. Before launch, and while running
  2. Roles
  3. Launch and observe
  4. Adapters and judgment
  5. Per-packet result
Original SKILL.md headings · 162 lines, 1,366 words
SKILL.md preview
---
name: agent-native
description: 'Dispatch independent tasks to parallel workers or subagents without write collisions. Use when: running or planning agents in parallel, even two; check scopes before any launch.'
practices: [team-topologies, design-by-contract]
hexagonal_role: supporting
consumes: [explicit-role-packets]
produces: [runtime-evidence, worker-handoff, per-packet-results]
context_rel:
- kind: customer-of
  with: codex-exec
- kind: customer-of
  with: claude-exec
skill_api_version: 1
user-invocable: true
metadata:
  tier: meta
  dependencies: []
  capabilities: [role_dispatch, observe_workers, handoff, dispatch_once]
  effects: [manage_runtime_sessions, invoke_selected_executor]
  canonical_status: canonical
  disposition: keep_optional_adapter
output_contract: runtime evidence and per-packet candidate, evidence, or error for explicit authorized work
---

# Agent Native

Launch and observe caller-selected agent sessions as explicit roles, and return
runtime facts per packet. Execution does not validate output, and the runtime
never becomes AgentOps lifecycle authority: an adapter cannot select AgentOps
semantics, issue a binding verdict, or turn factory completion into delivery or
validation proof. [Orchestrate](../orchestrate/SKILL.md) decides what to
dispatch and when; this skill launches and observes.

## Before launch, and while running

- **Validate the whole batch first.** Every packet needs its selected executor,
  packet identity, all transitive effects, the actions it must not take (such
  as push, merge, history rewrite, tracker or external writes) and a canonical
  workspace-relative write scope in separate isolation. If any packet is
  invalid, launch none: starting the valid ones and fixing the rest later is a
  partial launch.
- **Compare canonical scopes.** Resolve symlinks, normalize paths and compare
  case-insensitively so an alias cannot hide a collision. A lexical check alone
  cannot prove symlink or runtime isolation. Disjoint files are not
  independence: workers, readers included, contend on a resource one of them
  writes, installs into or holds exclusively, such as a bound port, a database
  it writes, a package install it mutates or a locked browser profile; give
  each its own instance or serialize only those. A shared model-provider limit
  is backpressure, launched in waves, not serialized
  ([shared runtime](references/launch-and-isolation.md#shared-runtime)).
- **Delivery is not engagement.** A delivered prompt or an acknowledgement
  proves transport only; reading it as a working worker is prompt-send
  optimism. Prove engagement from observable state: session output, tool
  activity, changed files.
- **Capture before restart.** Before a nudge, replacement or restart, capture
  the worker's observable state. A restart destroys the evidence of why it
  stalled, and rescue is usually cheaper than rerun. Silence or impatience alone
  does not justify a restart; any replacement stays within authority and
  remaining bounds.
- **Dispatch once.** Keep each packet's identity with its result. An executor
  error is a result, not a retry trigger; repair and follow-up are the caller's
  authority. A launch call the runtime refuses, so that its worker never
  starts, is not a launch: correct a malformed call once, and treat a
  concurrency or capacity limit as backpressure, waiting natively for a slot
  and launching the validated batch in waves under it; what each freed slot
  runs first stays the caller's decision
  ([Orchestrate](../orchestrate/SKILL.md#before-any-dispatch)). Running a
  packet inline or elsewhere is the caller's disclosed choice
  ([refused launches](references/launch-and-isolation.md#refused-launches)).
- **Account for every input.** When a packet carries many items (files,
  records, issues), give each a stable ID before launch and join the returned
  results back to that list. An item with no result is missing, not done; an
  output with no matching input is unexplained. Report both by ID. Name each
  return field and its allowed values in the packet when results will be
  counted, sorted, deduplicated or compared; parsed prose is where counts drift.
- **Prove integration before removal.** Before a caller-authorized removal of
  a finished worker's workspace, branch or stash, compare its content with the
  integration target as it stands now, by patch or file diff, not by merged
  status or by matching the commit that was merged: squash and rebase hide
  ancestry. Preserve uncommitted, unique or unintegrated content in
  a recoverable form, such as an archive ref or bundle, first. Report what was
  kept, what was removed and the evidence for each.

## Roles

- **Orchestrator:** passes focused intent, scope and evidence references, names
  the integration/final-review owner, and reports runtime facts. Retrieve extra
  history only for a consequential uncertainty; a fresh context is not
  necessarily small. A new goal does not clear history or renew spent bounds.
- **Implementer:** may modify only its packet's declared subject.
- **Validator:** receives exact candidate content in a fresh, read-only context.
  It may supply judgment to Validate; only Validate writes `verdict.v2`.
- **Scribe:** records runtime evidence without judging acceptance.

Reader and Writer are bounded cheap delegations, not roles with authority. A
Reader returns line-referenced bullets over files the caller never loads, in
slices of at most 350 lines. A Writer lands one patterned file from a spec plus
a required reference file and returns a receipt the caller never reads back.
Both are caller-selected per call and yield runtime facts only; a receipt is not
validation. For Codex, use the source-owned `bulk-reader` and `code-writer`
native roles; their model and sandbox are pinned in
[agents/bulk-reader.toml](agents/bulk-reader.toml) and
[agents/code-writer.toml](agents/code-writer.toml).
[Context-budget delegation](references/context-budget-delegation.md) covers
installation, native invocation, opt-in refusal hooks and the limits of role
instructions.

## Launch and observe

1. Before starting a worker, require caller intent, role, workspace, starting
   commit, authorized source/output scope and evidence destination. Record the
   dispatch association in the caller-owned native channel before execution can
   fail. A requested ID is not an observed ID; worker identity stays unknown
   until the runtime reports it.
2. At startup, before substantive work, capture the observed runtime, session
   and context identity and return it through that channel. Before its first
   edit, the worker confirms its HEAD is that starting commit; a mismatch is a
   launch failure, never repaired by reset. Report launch failures, recording
   gaps and unknowns; never fill them in. An authentication-shaped error from a
   sandboxed launch does not prove a logged-out account
   ([launch and isolation](references/launch-and-isolation.md#authentication-shaped-errors)).
   [Session associations](references/session-associations.md#work-to-session-associations)
   owns parent and resume links, multi-work spans and frozen source bounds.
3. Keep concurrent writers disjoint and isolated, using the selected runtime's
   own isolation where it has one and never nesting a worktree inside the
   session's own ([workspace isolation](references/launch-and-isolation.md#workspace-isolation)).
   Runtime coordination is not a claim, lease, queue or completion state in
   AgentOps.
4. Observe through native waits or status notifications; observation costs time
   and context. Unchanged state is no reason for another analysis, review or
   retrospective, while a known blocking failure deserves action even as other
   jobs run. Stop at terminal status or the end of the caller's window.
5. A headless worker cannot ask: on a choice its packet leaves open it takes no
   irreversible or out-of-scope step on a guess and returns the choice under
   `gaps` ([open choices](references/launch-and-isolation.md#open-choices)).
   For new authorized work after a worker completes, including an answer that
   settled intent supplies, use the runtime's documented follow-up or resume
   operation that starts a turn. A message operation only queues text for a
   running worker; a queued repair request is not a resumed attempt.
6. Record provider state, transcript references, artifacts and terminal status.
   Provider retries, reconnects, idle states and failures stay runtime facts,
   never Plan, Candidate or verdict state. A result whose observed tool
   activity shows an excluded action is out of authority, not a candidate,
   whatever its checks show.

The batch contract's reference implementation, `scripts/swarm/dispatch_once.py`,
needs an AgentOps source checkout and is not bundled with installed skills;
installed use dispatches through the selected native runtime. It rejects a
nonempty `write_scope.exclude` because its proof cannot honor exclusions. Batch
mode selects no backlog work, creates no queue and integrates no changes.

## Adapters and judgment

NTM, native processes, Agent Mail, Gas City and the one-shot headless runners
([codex-exec](../codex-exec/SKILL.md), [claude-exec](../claude-exec/SKILL.md),
[agy-native](../agy-native/SKILL.md)) are replaceable adapters. Use one only
when the caller selected that execution shape; a single local agent pays no
factory coordination cost.

For judgment, default to a fresh context in the author's model family.
Cross-model Validate, mixed Council and dueling model perspectives are explicit
caller selections. [Model dispatch](references/model-dispatch.md) owns host
authorization, isolation, finite input/output, timeout and cleanup for each leg;
the working session is the controller, no factory is required, and Agent Mail
is never the judgment path. Role requests declare authority, but only native
runtime/OS filesystem and egress controls enforce it.
[Native judgment receipts](references/judgment-receipts.md) define receipt
references and the checks for caller-required model diversity; missing native
identity never satisfies a leg. For exact native source spans, follow
[bounded raw source reads](references/RAW_SOURCE_READS.md) and its access and
output limits.

## Per-packet result

```text
packet_id:        the caller's packet id
executor:         selected runtime; requested model; observed model if reported
observed_session: session/context id the runtime reported, else unknown
engagement:       observable evidence the worker started, else unobserved
last_activity:    time and kind of the last observed tool call, output or file change, else unobserved
terminal_status:  exit or terminal state | running | launch failed | not launched
result:           candidate <ref> | evidence <ref> | out of authority <action> | error <verbatim>
gaps:             recording failures and unknowns
```
Files in this package (10)
  • SKILL.md 11 KB
  • agents/bulk-reader.toml 2 KB
  • agents/code-writer.toml 2 KB
  • references/RAW_SOURCE_READS.md 6 KB
  • references/context-budget-delegation.md 10 KB
  • references/judgment-receipts.md 6 KB
  • references/launch-and-isolation.md 5 KB
  • references/model-dispatch.md 11 KB
  • references/session-associations.md 5 KB
  • scripts/fake_model_runner.py 8 KB

Versions

The database has no other skills named agent-native.

Identical content appears in 1 other GitHub folders. We show one copy with the highest score.

Database snapshot: 13.11.2026 · package version 3249eef71c31 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 90partition-scopes-to-subagents
    Use when asked to partition non-overlapping scopes across subagents and drive each to completion. Not for concurrent independent sub-tasks: use parallel-launch.
  • 100loom-dispatching-parallel-agents
    Dispatch multiple independent tasks to parallel subagents when no shared file conflicts exist. Use when: a confirmed plan has independent tasks that can run concurrently without touching shared files.
  • 90parallel-agent-dispatch
    Dispatches independent subagents in parallel with clear contracts and merge discipline. Use this skill when multiple independent research or implementation strands can proceed without shared mutable state. Do not use when/for sequential dependent steps, or creating git worktrees…
  • 90parallel-launch
    Use when work splits into independent sub-tasks or cross-domain research. Not for durable scope ownership with per-scope review: use partition-scopes-to-subagents.
  • 100swarm
    Use to dispatch multiple parallel subagents on independent tasks. Discipline around isolation, conflict resolution, merge strategy. Wraps native parallel Task dispatch with structured gates.
  • 90trove-dispatch
    Parallel-agent dispatch workflow. Use when a request contains multiple independent tasks that can be safely delegated.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.