intake

A session-start process that clarifies a substantial request, checks for existing plans, anticipates follow-up work, and prepares one plan for approval before dispatching tasks. HITL means human-in-the-loop: the person is normally asked only for clarification and approval.

In plain words
What is it for?
It is for starting multi-step tasks, reusing existing plans, organising work for dispatch, and delivering results either in chat or as a rendered-page report.
Why use it?
It reduces repeated back-and-forth by settling the work and likely next steps at the beginning of a session.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/markmhendrickson/ateles/intake
Any agent
npx skills add markmhendrickson/ateles --skill intake
Clone the repo
git clone --depth 1 https://github.com/markmhendrickson/ateles

Made for: Claude Code, Codex.

Per session 0 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,627 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.01627
Opus 5 $0.00000 $0.00813
Sonnet 5 $0.00000 $0.00325
Haiku 4.5 $0.00000 $0.00163

Measured 2d ago against content hash 1cca8248421d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

intake scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/skills/intake/SKILL.md · 72 lines

How it starts

The opening of the file, as written. The whole thing — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.


name: intake description: "Fast session intake for the swarm leader: settle the plan/task with one clarification batch, cross-check and reuse existing plans/tasks, anticipate ALL follow-up work up front, then a single approval gate that dispatches the batch and delivers either an inline report or a rendered-page report emailed to the operator and linked for posterity. Collapses human-in-the-loop to (at most) one question batch + one approval. Auto-engages on the first substantive request of a session; also invocable as /intake." slug: intake user_invocable: true triggers:

  • intake
  • /intake
  • start session
  • what should we do
  • new request supported_harnesses:
  • claude-code
  • cursor entity_id: ent_a650ca318a06a92e54bee312

intake

The swarm leader's session-intake protocol. Goal: from the operator's opening request, reach a dispatched plan with the fewest possible human-in-the-loop turns — ideally one clarification batch and one approval gate — then deliver results the way the operator chose.

When to engage

  • Auto-engage on the first substantive request of a session (a task or goal), not on quick lookups or one-step asks you can just do.
  • Or invoke explicitly: /intake <request>.
  • Skip entirely for trivial single-step work — do it directly.

Phase 1 — Settle the work (ONE clarification pass)

  1. Bounded retrieval. retrieve_entities over plan and task matching the request (lexical + semantic). Search by CONCEPT and architecture terms, not only the product/feature noun — run several phrasings (the capability or "spine/loop/lifecycle/pipeline" name, the owning role, the core verb), because the best-matching plan is frequently named for the concept rather than the product. A single product-noun query is how a concept-named plan gets missed and a duplicate, competing plan gets created (the cross-workstream-collision failure mode).
  2. Cross-check, don't duplicate. Decide whether an existing plan/task already fulfills the request — fully, or better than asked. If so, propose executing/refining THAT (cite entity IDs) instead of creating something new.
  3. Bind the plan. Pick the closest existing plan as the session's bound plan; if none fits, create one (/update-plan). Never write one workstream's items into another workstream's plan.
  4. Ask all forks at once. Put every genuine decision fork into a SINGLE AskUserQuestion batch; infer sensible defaults for everything else. Do not drip questions across turns.

Read the full file on GitHub · 72 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 72 lines · 0 tokens per session scan A 1cca8248421d

Subscribe to this mod's changes

intake is a skill published in the GitHub repository markmhendrickson/ateles (5 stars, last pushed 5d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,627 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

10x-orchestrator

MANDATORY - The core brain of 10X Vibe Marketer. You are the autonomous CMO orchestrating a full marketing agency of 12 specialist departments. Load this skill for ANY marketing task - content creation, SEO, campaigns, strategy, research, analytics, outreach, social media, email, paid ads, branding, funnel…

OpenAnalystInc/Vibe-Marketer · 91 tokens

10x-setup

First-time setup and initialization for 10X Vibe Marketer. Includes an interactive wizard that lets users choose their database, features, marketing focus, visual tools, and tracking preferences. Checks setup.log to determine if setup has already been completed. Run via /setup or auto-triggered on first use.

OpenAnalystInc/Vibe-Marketer · 66 tokens

10x-visualizer

Generate visual marketing deliverables — strategy boards, funnel diagrams, flowcharts, journey maps, dashboards, and more — using TLDraw SDK (local .tldr files). Each department agent creates visuals for their own domain. Invoke with /report.

OpenAnalystInc/Vibe-Marketer · 55 tokens

community-manager

Skill "community-manager" from OpenAnalystInc/Vibe-Marketer, covering community manager — 10x vibe marketer, identity, step-by-step execution, cli visual identity and expertise.

OpenAnalystInc/Vibe-Marketer · 0 tokens

10x-research

The research orchestration engine for 10X Vibe Marketer. Powers multi-platform, real-time research across 13 platforms using Claude in Chrome and WebSearch/WebFetch. Spawns platform-specific research agents in parallel, consolidates findings through the research lead, and delivers actionable marketing intelligence.

OpenAnalystInc/Vibe-Marketer · 0 tokens

affiliate-specialist

Skill "affiliate-specialist" from OpenAnalystInc/Vibe-Marketer, covering affiliate specialist — 10x vibe marketer, identity, step-by-step execution, cli visual identity and expertise.

OpenAnalystInc/Vibe-Marketer · 0 tokens