Skip to content
Back to skills

Ingest

BSecurity

Bootstraps the memory store from existing material (repos, PDFs, transcripts). Use to seed memory from a project or document.

  • 30 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added October 3, 2026
ai-agentspythongoshellbashgitapi

Works with

  • cli
  • api
  • mcp

Security analysis

B88/100
  • criticalDownloads and executes remote scripts — classic supply chain attack

Pro shows the line behind each finding and how to fix it

Scanned October 3, 2026

npx -y skills add fjwood69/mori --skill ingest --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Ingest?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Ingest
[![Security: B — Skills Directory](https://www.skillsdirectory.com/api/skills/fjwood69-ingest/badge)](https://www.skillsdirectory.com/skills/fjwood69-ingest)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: ingest
description: Bootstraps the memory store from existing material (repos, PDFs, transcripts). Use to seed memory from a project or document.
---

## Filesystem mode (server-local paths)

1. If the user passes `--status`, call `mori-ingest_status` and present the table.
2. If the user passes `--preview` or `--dry-run` with no source provided, remind them to provide `--source <path>`.
3. For `--preview`: call `mori-ingest_preview` with the `--source` path(s) and any `--type` or `--since` arguments. Present the chunk breakdown and cost estimate. Remind the user this is zero-cost — no LLM was called.
4. For `--dry-run`: call `mori-ingest` with `dry_run=true` and all provided arguments. Report what would be written and the actual cost incurred (the LLM was called, just nothing committed).
5. For a real ingestion run: call `mori-ingest` with all provided arguments. Report: sources processed, chunks sent, memories written, cost estimate, errors.
6. If errors occurred for specific files, report them individually.
7. After successful ingestion, suggest running `/brief` to reload shared memories so the new entries are visible.
8. For large or expensive-looking sources, suggest `/ingest --preview` first to check before committing.

## Content mode (remote clients — HTTP upload)

When the user passes `--content <path>`, upload the file directly to the mori-ingestion pod.
Do NOT use the `mori-ingest_content` MCP tool — it has been removed. Use HTTP multipart upload only.

Requires:
- `MORI_INGEST_URL` env var (set by installer — NOT derived from `MORI_URL`)
- `MORI_API_KEY` env var (same key as MCP auth)

### Linux/macOS

```bash
JOB=$(curl -sf -X POST "${MORI_INGEST_URL}/api/ingest/upload" \
  -H "X-Api-Key: ${MORI_API_KEY}" \
  -F "files=@${path}" \
  -F "focus=${focus}" \
  -F "tier=${tier}" \
  -F "tags=${tags}" \
  -F "dry_run=${dry_run}")
JOB_ID=$(echo "$JOB" | python3 -c "import sys,json; print(json.load(sys.stdin)['job_id'])")
```

### Windows (PowerShell)

```powershell
$r = Invoke-RestMethod -Uri "$env:MORI_INGEST_URL/api/ingest/upload" `
  -Method POST `
  -Headers @{"X-Api-Key" = $env:MORI_API_KEY} `
  -Form @{
    files   = Get-Item $path
    focus   = $focus
    tier    = $tier
    tags    = $tags
    dry_run = $dry_run
  }
$JOB_ID = $r.job_id
```

### Polling (both platforms)

Poll every 5 seconds until `status` is `complete` or `failed`:

```bash
# Linux/macOS
while true; do
  STATUS=$(curl -sf "${MORI_INGEST_URL}/api/ingest/job/${JOB_ID}" \
    -H "X-Api-Key: ${MORI_API_KEY}")
  STATE=$(echo "$STATUS" | python3 -c "import sys,json; print(json.load(sys.stdin)['status'])")
  [ "$STATE" = "complete" ] || [ "$STATE" = "failed" ] && break
  sleep 5
done
echo "$STATUS"
```

```powershell
# Windows
do {
  $status = Invoke-RestMethod -Uri "$env:MORI_INGEST_URL/api/ingest/job/$JOB_ID" `
    -Headers @{"X-Api-Key" = $env:MORI_API_KEY}
  if ($status.status -notin "queued","running") { break }
  Start-Sleep 5
} while ($true)
$status
```

Report: memories written, estimated cost, errors. After success, suggest `/brief` to reload shared memories.

### Error handling

- `MORI_INGEST_URL` not set → tell user to set it (e.g. `http://localhost:8969`) and re-run
- HTTP 413 → file too large (50MB/file, 200MB total)
- HTTP 401 → API key mismatch — check `MORI_API_KEY`
- status `failed` → report `errors` field from job response

## Argument mapping

| User flag | Action |
|-----------|--------|
| `--source <path>` (repeatable) | `mori-ingest` MCP tool → `source` list |
| `--content <path>` | HTTP multipart upload to `MORI_INGEST_URL` |
| `--type <type>` | `type` (auto, transcripts, git, docs, image) |
| `--focus <area>` | `focus` (all, decisions, architecture, conventions, gotchas) |
| `--tier <tier>` | `tier` (working, canonical, ephemeral) |
| `--tags <tags>` | `tags` (comma-separated string) |
| `--since <duration>` | `since` (e.g. "30d", "90d") — filesystem mode only |
| `--dry-run` | `dry_run=true` |
| `--force` | `force=true` |
| `--max-cost <amount>` | `max_cost` (float, USD) — filesystem mode only |
| `--preview` | call `mori-ingest_preview` instead of `mori-ingest` |
| `--status` | call `mori-ingest_status` |

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.