fdeops 3.23.0 → 3.26.0

package/README.md

@@ -1,12 +1,10 @@
 # FDEOps
 
-**Forward deployed engineering skills for AI coding agents.**
+**One client record, from first meeting to handover.**
 
-You're on a customer site. The AI coding agent writes code in their repo. This kit is the work around that code: the brief, who can say yes, proof on their staging then live, whether they signed off, whether they can run it after you leave.
+FDEOps helps a Forward Deployed Engineer or independent expert manage client work with an AI coding agent. Keep the brief, decisions, delivery evidence, acceptance, and next actions together across repositories and sessions.
 
-Notes stay on your laptop. Their repo stays theirs. You confirm before anything is written down.
-
-Keep the coding pack you already use. Install this next to it. FDEOps is the client work. The other pack writes the code.
+One `@fde` skill guides the work. A local CLI records it in Markdown. A read-only dashboard shows one client or your portfolio. You review judgments and obtain customer approval; the tool does not decide for them.
 
 <img width="1536" height="1024" alt="fdeops" src="https://github.com/user-attachments/assets/2bcb8739-55ee-445d-8a1a-8b38433b7b58" />
 
@@ -14,13 +12,13 @@ Keep the coding pack you already use. Install this next to it. FDEOps is the cli
 
 ## Quick Start
 
-**Try it first, nothing installed.** In any repo:
+**Inspect a repository first.** Requires Node.js 18+ and Git. Run from a local repository:
 
 ```bash
 npx fdeops scan
 ```
 
-Two minutes. It prints what to look at on day one and the questions to ask. Local only, nothing written.
+`npx` may download the package. The scan itself reads local files, prints reconnaissance and questions, and does not write engagement records.
 
 **Then install the skill:**
 
@@ -34,9 +32,15 @@ One chat. Name the client:
 @fde this is client01
 ```
 
-That creates `~/fde-engagements/client01/.fde/` on your laptop. Paste kickoff notes in the same thread. `@fde` picks what to check. You still decide. After a meeting you get one screen: what changed, new asks, open questions, next actions. Confirm once.
+The agent creates and binds `~/fde-engagements/client01/.fde/` to the workspace. Paste kickoff notes; review the proposed decisions, risks, and next actions before saving. If setup cannot run in your host, use `npx fdeops resume --init client01` to create the engagement and binding.
+
+Open the engagement fieldbook anytime:
 
-Day to day: [docs/USAGE.md](docs/USAGE.md).
+```bash
+npx fdeops dashboard --open
+```
+
+Read-only HTML of the record - promised, measured, accepted, and evidence. Regenerate after you change memory. Day to day: [docs/USAGE.md](docs/USAGE.md).
 
 <details>
 <summary><b>Claude Code</b></summary>
@@ -46,7 +50,7 @@ Day to day: [docs/USAGE.md](docs/USAGE.md).
 /plugin install fdeops@fdeops
 ```
 
-Hooks resume where you left off. Slash commands match the table below.
+The plugin registers Claude Code session hooks and slash commands. A skill-only install does not register hooks.
 
 </details>
 
@@ -64,10 +68,12 @@ See [adapters/](adapters/README.md).
 </details>
 
 <details>
-<summary><b>Air-gap, PATH, override</b></summary>
+<summary><b>Clone install and offline use</b></summary>
 
 ```bash
-git clone https://github.com/suboss87/fdeops.git && node bin/install.js
+git clone https://github.com/suboss87/fdeops.git
+cd fdeops
+node bin/install.js
 ```
 
 If the agent cannot create the folder:
@@ -76,7 +82,7 @@ If the agent cannot create the folder:
 npx fdeops resume --init client01   # ~/fde-engagements/client01
 ```
 
-Requires Node.js >= 18. Override: `FDEOPS_ENGAGEMENT`. See [docs/install.md](docs/install.md). Try the loop: `npx fdeops demo`.
+For offline use, transfer the checkout first; cloning and package downloads need network access. Override: `FDEOPS_ENGAGEMENT`. See [docs/install.md](docs/install.md).
 
 </details>
 
@@ -86,7 +92,7 @@ Requires Node.js >= 18. Override: `FDEOPS_ENGAGEMENT`. See [docs/install.md](doc
 
 One command per stage. Skills load automatically.
 
-Six stages, same order every job: Land, Discover, Plan, Ship, Outcome, Close.
+Six stages organize the engagement: Land, Discover, Plan, Ship, Outcome, Close. Start with the current situation; revisit earlier decisions when the evidence changes. Slash commands below are provided by the Claude Code plugin. In other hosts, describe the same request to `@fde`.
 
 | What you're doing | Command | Stage |
 |-------------------|---------|-------|
@@ -99,79 +105,36 @@ Six stages, same order every job: Land, Discover, Plan, Ship, Outcome, Close.
 
 Same `@fde`, when you need them: `/debrief` (notes into the record), `/prep` (one page before you walk in), `/trust` (process gap, or they stopped trusting you), `/receipts` (a dated line, or it did not happen), `/readout` (Friday page for the sponsor; not a seventh stage).
 
-You can also just say it: naming a client, a POC, changing their checkout, going live, asking what was agreed. A typo in a repo that is not a client job can skip this. A named client, a POC, or go-live cannot.
+You can also describe the situation in plain English: a new client, a POC, an incident, a scope change, or a question about what was agreed.
 
 ---
 
-## All 30 Skills
-
-Thirty situations, grouped by stage. Not prompts - each one has steps, a file it writes, and a checkpoint with you. Type English or a slash command. `@fde` opens the matching skill. You never pick one by name.
+## Your daily fieldbook
 
-Full detail: [docs/skills-reference.md](docs/skills-reference.md).
+![FDEOps dark dashboard showing next actions and attention gaps across three fictional clients](media/fieldbook-preview.png)
 
-### Land
+*Fictional client records, shown in the built-in dark theme. The report works offline.*
 
-| Skill | What it does | Use when |
-|--------|--------------|----------|
-| [land](skills/fde/references/land.md) | Interrogate the brief | New client, first meeting, just got the brief |
-| [audit](skills/fde/references/audit.md) | Verify inherited claims | Taking over, previous consultant left |
-| [who-decides](skills/fde/references/who-decides.md) | Map decision rights | Need to know who matters |
-| [earn-trust](skills/fde/references/earn-trust.md) | Earn access | Need access or credibility |
-| [hold-scope](skills/fde/references/hold-scope.md) | Hold scope | "Also can you…", timeline unchanged |
+Read [verification results and limits](docs/verification.md) for context measurements, local-model observations, and MCP coverage.
 
-### Discover
+Open `npx fdeops dashboard --all --open` to review every client. Filter what needs attention, open a client, and copy **Continue next action** into your agent. After a meeting, use **Debrief notes**; before a sponsor update, use **Review outcome**. These buttons copy prompts; work runs in your agent. Review changes there and regenerate the dashboard afterward.
 
-| Skill | What it does | Use when |
-|--------|--------------|----------|
-| [discover](skills/fde/references/discover.md) | Frame the problem | Brief feels wrong, shadow processes |
-| [test-assumptions](skills/fde/references/test-assumptions.md) | Test assumptions | Brief feels too neat |
-| [score-use-cases](skills/fde/references/score-use-cases.md) | Score use cases | Everything is P0 |
-| [poc](skills/fde/references/poc.md) | Validate the solution | POC, spike, need to de-risk |
+## Try the complete loop
 
-### Plan
-
-| Skill | What it does | Use when |
-|--------|--------------|----------|
-| [plan](skills/fde/references/plan.md) | Sequence the work | What order, what is done |
-| [business-case](skills/fde/references/business-case.md) | Build the business case | Defend budget or timeline |
-| [three-options](skills/fde/references/three-options.md) | Generate options | "What should we do?" |
-| [pick-three](skills/fde/references/pick-three.md) | Prioritize three | Everything is urgent |
-
-### Ship
-
-| Skill | What it does | Use when |
-|--------|--------------|----------|
-| [ship](skills/fde/references/ship.md) | Deliver the increment | Building, updating, or going live |
-| [what-breaks](skills/fde/references/what-breaks.md) | Assess impact | Touching shared infrastructure |
-| [rescue](skills/fde/references/rescue.md) | Resolve the incident | Down, or they went quiet |
-| [review](skills/fde/references/review.md) | Review the change | Before merge, scope creep |
-| [rollback](skills/fde/references/rollback.md) | Rehearse rollback | "We can always revert" |
+```bash
+npx fdeops demo
+```
 
-### Outcome
+This writes fictional records and HTML under `~/fde-engagements/.demo/`, resetting its own sandbox on each run. It requires no AI account. Open the generated report; remove the demo later with `npx fdeops demo --clean`. Follow the [five-minute walkthrough](docs/USAGE.md#new-here-5-minutes) for what to inspect.
 
-| Skill | What it does | Use when |
-|--------|--------------|----------|
-| [readout](skills/fde/references/readout.md) | Report the outcome | Friday, sponsor update |
-| [demo-prep](skills/fde/references/demo-prep.md) | Prepare the demo | Demo or exec walkthrough |
-| [debrief](skills/fde/references/debrief.md) | Capture the meeting | Just left a meeting |
-| [board-memo](skills/fde/references/board-memo.md) | Brief the board | Justify continued investment |
-| [dashboard](skills/fde/references/dashboard.md) | View the portfolio | All my customers |
-| [ingest](skills/fde/references/ingest.md) | Ingest sources | Transcript, Notion, Slack |
-| [connect](skills/fde/references/connect.md) | Connect a source | Connect Granola |
 
-### Close
+---
 
-| Skill | What it does | Use when |
-|--------|--------------|----------|
-| [close](skills/fde/references/close.md) | Transfer operations | Wrapping up |
-| [runbook](skills/fde/references/runbook.md) | Write the runbook | They must operate without you |
-| [switch-clients](skills/fde/references/switch-clients.md) | Switch engagements | 2+ clients |
-| [encode-pattern](skills/fde/references/encode-pattern.md) | Encode the pattern | It will apply again |
-| [red-team](skills/fde/references/red-team.md) | Challenge the plan | "Poke holes in this" |
+## All 30 Skills
 
-Overlays (on signal, not on request): [ai](skills/fde/references/ai.md) · [artifacts](skills/fde/references/artifacts.md) · [fintech](skills/fde/references/fintech.md) · [healthcare](skills/fde/references/healthcare.md) · [gov](skills/fde/references/gov.md). AI companion (not a sixth overlay): [eval-pack](skills/fde/references/eval-pack.md).
+Not prompts to choose from: the router loads one relevant skill with concrete steps, an artifact, and a checkpoint. The [skills reference](docs/skills-reference.md) lists all 30, from discovery and scope control to incident response and handover. Industry and AI overlays apply when relevant.
 
-Optional pull: you add the source MCP; we **pull** on request. [mcp/recipes/](mcp/recipes/)
+Optional external sources use your configured MCP connections. The agent pulls on request, stages the material, and asks you to review its interpretation before applying it. The CLI itself stays local. See [source recipes](mcp/recipes/) and [ingest usage](docs/USAGE.md#ingest-pull-large-artifacts--same-confirm-loop).
 
 ---
 
@@ -191,13 +154,16 @@ One `@fde`. One file per situation. One folder per client.
            │  after you confirm
            ▼
   ~/fde-engagements/client01/.fde/
+           │
+           ▼
+  fde dashboard --open         offline fieldbook (read-only snapshot)
 ```
 
-**A dated line, or it did not happen.** Promised → measured → accepted. If it is not in `.fde/`, it is not on the record.
+**A dated line, or it did not happen.** Promised → measured → accepted, with evidence for the measurement. If it is not in `.fde/`, it is not on the record.
 
-**Confirm, then it is written.** The CLI stays on your laptop: git and files, no network. The AI coding agent runs the command. You say yes. Then it is in the folder.
+**Review judgments before recording them.** The agent shows proposed interpretations for confirmation. Direct CLI writes execute when invoked; enabled session hooks also capture mechanical session state automatically. Recording a note does not establish customer approval.
 
-**The record is on your laptop.** Change hosts, install `@fde` on the new one, keep talking. The notes are not inside any vendor.
+**The record is on your laptop.** Change hosts and keep the same Markdown. Claude Code plugins provide automatic session hooks; other hosts use `@fde` and CLI commands on demand. See the [install matrix](docs/install.md#who-needs-which-install).
 
 One skill hosts load: `skills/fde/SKILL.md`. It opens one file in `skills/fde/references/` and stops. Slash commands live in `.claude/commands/`. The local CLI is `bin/fde.js` (git + files, no network). Layout: [docs/REPO_LAYOUT.md](docs/REPO_LAYOUT.md).
 
@@ -212,17 +178,17 @@ One folder per client. Plain markdown. Grep it, copy it, take it into a meeting.
 | `context.md` | Where you are |
 | `brief.md` / `success.md` | What they asked; what “done” is and who signs |
 | `reality.md` / `terrain.md` | The real problem; the map |
-| `stakeholders.md` | `[signal:green\|amber\|red]` |
+| `stakeholders.md` | `[signal:green\|amber\|red]` - worst active signal wins; empty is **new**, not green |
 | `trust-profile.md` | Sacred data, AI policy, approval chain |
-| `decisions.md` / `risks.md` / `delivery.md` | Dated choices; live risks; what shipped and how it rolls back |
+| `decisions.md` / `risks.md` / `delivery.md` | Dated choices; live risks; what shipped, evidence, rollback, acceptance |
 
-Schema: [docs/schema.md](docs/schema.md). Local HTML: `npx fdeops dashboard`.
+Schema: [docs/schema.md](docs/schema.md). Fieldbook: `npx fdeops dashboard --open` (bound) or `--all --open` (portfolio).
 
 ---
 
 ## Who this is for
 
-You sit with a customer's team. An AI coding agent writes in their repo. You need a record of the brief, who can say yes, what went live, and whether they signed off.
+Forward Deployed Engineers, independent consultants, and solo agencies delivering inside customer systems. Use a separate engagement folder for each client and a workspace binding to select the right record.
 
 If you ship your own company's product from HQ, with no customer team that has to run it after you leave, you do not need this kit.
 
@@ -230,99 +196,40 @@ If you ship your own company's product from HQ, with no customer team that has t
 
 ## Your data stays yours
 
-The **CLI** is local: git + files, no network, no telemetry. The **host model** sees `.fde/` the agent loads (usually a bounded `context.md`) and any client code you open. It must not see `<private>` blocks - redacted from CLI, dashboard, and hooks; do not paste them or open them with file tools. Nothing is written until you confirm. `~/fde-engagements` is in `$HOME`; iCloud/Dropbox is an NDA incident waiting.
+The **CLI** is local: git + files, no network, no telemetry. The **host model** sees `.fde/` the agent loads (usually a bounded `context.md`) and any client code you open. It must not see `<private>` blocks - redacted from CLI, dashboard, and hooks; do not paste them or open them with file tools. Keep the engagement folder outside cloud-sync locations unless customer policy permits that storage. Generated reports can still contain confidential information after redaction; review them before sharing.
 
 [PRIVACY.md](PRIVACY.md) · [SECURITY.md](SECURITY.md)
 
 ---
 
-## Why FDEOps?
-
-AI coding agents are built for a repo, not for a client. Left alone they skip who signs, whether the brief is true, and whether anyone accepted the number. Monday they start from the ticket again.
-
-This is the kit you take on site. `@fde` runs the client work around the code. A local command dates every decision. The notes are markdown on your laptop. You confirm; then it is on the record.
-
----
-
 ## Principles
 
 - **Who signs** - name them in the first days
 - **Brief vs real job** - check the floor, not only the slide
 - **Back from done** - sequence from signed-off, not from the ticket
 - **Their staging then live** - prove it where they operate, then go live
-- **Promised, measured, accepted** - a number nobody signed is claimed, not delivered
+- **Promised, measured, accepted** - a number nobody signed is claimed, not delivered; keep the evidence
 - **They run it** - if they cannot operate it without you, you are not done
 - **A dated line, or it did not happen** - these files get defended in the room
-- **One customer, one folder** - context never bleeds
+- **One customer, one folder** - verify the active binding before writing
 - **The kit says what to check. You still decide.**
 
 ---
 
 ## Project Structure
 
-```
-fdeops/
-├── skills/fde/                            # the one skill hosts load
-│   ├── SKILL.md                           #   router
-│   └── references/                        #   30 skills + overlays (you never pick)
-│       ├── land.md                        #   Land
-│       ├── audit.md
-│       ├── who-decides.md
-│       ├── earn-trust.md
-│       ├── hold-scope.md
-│       ├── discover.md                    #   Discover
-│       ├── test-assumptions.md
-│       ├── score-use-cases.md
-│       ├── poc.md
-│       ├── plan.md                        #   Plan
-│       ├── business-case.md
-│       ├── three-options.md
-│       ├── pick-three.md
-│       ├── ship.md                        #   Ship
-│       ├── what-breaks.md
-│       ├── rescue.md
-│       ├── review.md
-│       ├── rollback.md
-│       ├── readout.md                      #   Outcome
-│       ├── demo-prep.md
-│       ├── debrief.md
-│       ├── board-memo.md
-│       ├── dashboard.md
-│       ├── ingest.md
-│       ├── connect.md
-│       ├── close.md                       #   Close
-│       ├── runbook.md
-│       ├── switch-clients.md
-│       ├── encode-pattern.md
-│       ├── red-team.md
-│       ├── ai.md                          #   overlays (on signal)
-│       ├── artifacts.md
-│       ├── eval-pack.md
-│       ├── fintech.md
-│       ├── healthcare.md
-│       └── gov.md
-├── .claude/commands/                      # slash commands (each loads @fde)
-│   ├── brief.md
-│   ├── discover.md
-│   ├── plan.md
-│   ├── ship.md
-│   ├── outcome.md
-│   ├── close.md
-│   ├── trust.md
-│   ├── receipts.md
-│   ├── debrief.md
-│   ├── prep.md
-│   └── readout.md
-├── .claude-plugin/                        # Claude Code marketplace
-├── bin/                                   # local CLI: git + files, no network
-├── hooks/                                 # session-start / session-stop / pre-compact
-├── adapters/                              # Cursor, Gemini, Copilot, Codex pointers
-├── templates/.fde/                        # memory files created on first client
-├── examples/                              # fictional walkthroughs
-├── mcp/                                   # optional ingest + source recipes
-├── evals/                                 # routing checks
-└── docs/                                  # usage, schema, install
-```
+| Folder | Responsibility |
+|---|---|
+| `skills/fde/` | One router and the delivery methodology |
+| `bin/` | Local CLI, installer, and shared helpers |
+| `templates/.fde/` | Client record templates |
+| `adapters/` and `hooks/` | Host entry points and session integration |
+| `mcp/` | Optional ingest wrapper and source recipes |
+| `test/` and `evals/` | Code regressions and workflow evaluations |
+| `docs/` and `examples/` | Usage, contributor guides, and fictional engagements |
+
+See [repository layout](docs/REPO_LAYOUT.md) for where to make changes.
+
 
 ---
 

package/adapters/LOCAL-LLM.md

@@ -1,110 +1,22 @@
-# fdeops - Local LLM Setup
+# Local models
 
-Use fdeops with **any local model** - Ollama, LM Studio, llama.cpp, vLLM, Open WebUI, or any inference server that accepts a system prompt.
+The FDEOps CLI and offline dashboard need Node.js and Git, not a model. AI-assisted work additionally needs a local model **and an agent host that can read files and run approved CLI commands**. A chat box with the skill pasted into its system prompt does not automatically gain those capabilities.
 
-## Why it works
+## Use your existing local agent
 
-fdeops is a SKILL.md file + markdown memory + a Node.js CLI. It calls no external API. The AI does the skills; the CLI does the mechanics. Any model that can read a markdown system prompt can run fdeops.
+1. Download FDEOps, your agent host and your model while online. After that, the CLI operates offline. Model/provider configuration belongs to the host; FDEOps does not start or configure an inference server.
+2. From the client workspace, run `node /path/to/fdeops/bin/fde.js resume --init my-client` to create and bind a local record.
+3. Make `skills/fde/SKILL.md` and its references available to the host. Use the host's documented skill/file mechanism. Give it the FDEOps CLI path and permission to read the bound record and execute the requested commands.
+4. Start with `fde resume` and ask for the next action. Inspect the tool calls, cited records and any proposed writes before trusting the workflow.
 
-## Setup
+Use `fde recall <topic>` for relevant evidence. `resume` and `recall` default to a 16 KiB output ceiling; `--max-bytes 4096` requests a smaller allowance. This limits FDEOps output, not the host's entire context window. Keep unrelated transcripts and tools out of the active context. Private blocks must stay out of direct file reads as well as prompts.
 
-### 1. Install fdeops (same as any other setup)
+## What compatibility means
 
-```bash
-npx fdeops init my-client
-```
+- **CLI verified:** commands work without a model.
+- **Transport verified:** the host can call a model or MCP server and receive a valid response.
+- **Workflow verified:** that exact host/model combination selects tools, uses the correct client, preserves constraints, and returns a useful result.
 
-### 2. Load the skill into your local model
+These are separate checks. Model size alone does not establish quality, and FDEOps does not promise support for every model. Record the model/version, context setting, tool permissions, task, and actual outcome. Test a scope change, missing evidence and a refused approval before relying on a new model for client delivery.
 
-The file your model needs to read as system context:
-
-```
-# If you cloned the repo:
-skills/fde/SKILL.md
-
-# If you ran `node bin/install.js` (also sets up hooks + adapters):
-~/.claude/skills/fde/SKILL.md
-```
-
-Either path works - same file. If you only want the local LLM workflow and skipped `install.js`, use the repo-local path directly.
-
-How you load it depends on your setup:
-
-| Tool | How to load |
-|------|-------------|
-| **Ollama + Open WebUI** | Paste the contents of `SKILL.md` into the system prompt field, or mount it as a file in your Modelfile |
-| **LM Studio** | Add `SKILL.md` path to the system prompt in chat settings |
-| **llama.cpp / server mode** | Pass `--system-prompt-file skills/fde/SKILL.md` |
-| **vLLM + chat UI** | Include as the system message in your chat template |
-| **Aider** | Run aider from your engagement workspace - it reads repo files including SKILL.md automatically |
-| **Continue.dev (VS Code)** | Add SKILL.md as a context provider in `.continue/config.json` |
-| **text-generation-webui** | Load SKILL.md content in the "Context" or "System prompt" tab |
-| **Jan.ai** | Paste into the system prompt field in assistant settings |
-| **GPT4All** | Add as system prompt in the model's chat configuration |
-
-### 3. Set the engagement path
-
-Your local model needs to know where the engagement memory lives. Set the environment variable before starting your session:
-
-```bash
-export FDEOPS_ENGAGEMENT=~/fde-engagements/my-client/.fde
-```
-
-Or tell the model directly: "My engagement is at ~/fde-engagements/my-client/.fde"
-
-### 4. Use it
-
-```
-@fde I'm preparing for tomorrow's stakeholder meeting. The brief says payments API.
-```
-
-The model reads SKILL.md, routes to the right skill, and produces artifacts in your `.fde/` folder.
-
-## Model size recommendations
-
-The kit is detailed (30 skills, routing logic, evidence format, memory contract). Larger models handle it better:
-
-| Model class | Experience |
-|-------------|-----------|
-| **7-8B** (Llama 3.1 8B, Mistral 7B, Qwen 2.5 7B) | Handles individual skills (readout, log, land). May struggle with complex routing or multi-skill sessions. Good for the CLI-heavy workflow where you invoke skills explicitly. |
-| **13-34B** (Llama 3.1 13B, Mixtral 8x7B, Qwen 2.5 32B, DeepSeek-R1 32B) | Good across all domains. Routes correctly, follows memory contract, writes structured artifacts. Recommended minimum for full use. |
-| **70B+** (Llama 3.1 405B, DeepSeek V3, Qwen 2.5 72B) | Full capability. Handles regulated overlays, switch-clients, board-memo pyramid, runbook handoff. |
-
-## The CLI works without ANY model
-
-Even if you can't run a local model (or you're at a regulated client with no AI permitted), the `fde` CLI gives you the deterministic tooling:
-
-```bash
-fde scan          # repo recon - hotspots, test gaps, AI components (git only)
-fde resume        # load/create engagement memory
-fde log           # structured append: decisions, risks, delivery, contacts
-fde receipts      # "what did we agree?" - search memory with dates
-fde status        # portfolio triage across all customers (red/amber/green)
-fde dashboard     # offline HTML fieldbook across all engagements
-fde capture       # session-end memory snapshot
-```
-
-Zero network. Zero AI. Pure local Node.js reading git and markdown.
-
-## Ollama Modelfile example
-
-```dockerfile
-FROM llama3.1:70b
-
-# System prompt provided at runtime via --system flag
-
-PARAMETER temperature 0.3
-PARAMETER num_ctx 32768
-```
-
-Then load the skill:
-```bash
-ollama run my-fde-model --system "$(cat skills/fde/SKILL.md)"
-```
-
-## Tips for local models
-
-- **Context window matters.** SKILL.md + references can be large. Use a model with at least 8K context; 32K+ is ideal for loading skill references on demand.
-- **Temperature 0.2-0.4 works best.** The skills are structured - lower temperature keeps routing accurate and artifacts consistent.
-- **Use the CLI for mechanics.** Don't ask the model to do what the CLI already does deterministically. Use `fde scan` for repo recon, `fde log` for memory writes, `fde receipts` for searching. Let the model handle judgment, routing, and artifact drafting.
-- **Explicit skill invocation.** If a smaller model struggles with routing, you can invoke skills directly: "@fde run the discover phase" or "@fde use hold-scope." The model skips routing and goes straight to the skill.
+See [verification results](../docs/verification.md) for the combinations actually exercised. No automatic Claude session hooks are implied for local agent hosts.

package/bin/check.js

@@ -449,9 +449,9 @@ const hookCode = hook.replace(/^[ \t]*#.*$/gm, '')
 if (!hook.includes('FDEOPS_ENGAGEMENT')) {
   fail('session-start hook must read FDEOPS_ENGAGEMENT env var')
 } else ok('hook FDEOPS_ENGAGEMENT')
-if (!hookCode.includes('FDEOPS_ENGAGEMENT="$ENG_DIR" fde triage') ||
-    !hookCode.includes('FDEOPS_ENGAGEMENT="$ENG_DIR" node "$FDE_CMD" triage')) {
-  fail('session-start must run triage with its resolved engagement')
+if (!hookCode.includes('FDEOPS_ENGAGEMENT="$ENG_DIR" fde resume') ||
+    !hookCode.includes('FDEOPS_ENGAGEMENT="$ENG_DIR" node "$FDE_CMD" resume')) {
+  fail('session-start must delegate bounded context and triage to resume with its resolved engagement')
 }
 // Token discipline: SessionStart must not dump the full skill (L1 progressive disclosure).
 // Strip comments before scanning for a real `cat …SKILL.md` / BOOTSTRAP inject.