Skip to content
Skills Library
AI & agents · Building & coding

develop-with-memory

/develop-with-memory

Use when you are investigating, fixing or changing code in a repository and your long-term memory is in Vornik: recall before reading code, design before changing it, keep memory current, and, where memory holds no design for the part being changed, write an as-built record of that part with the user, then store it. Never claims a design exists: it reads each recall's design_record count.

Repository
vornik ↗
Licence
AGPL-3.0
Last repository change
25.10.2026

Assessment and security

90/ 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.
Content25 / 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
1
Scripts
0
HTML comments
4
URLs in SKILL.md
0
Deprecation

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

Licence

AGPL-3.0, copyleft: przeróbki udostępniane innym muszą mieć tę samą licencję.

What to check before installing
  • SKILL.md zawiera komentarze HTML (4). Nie widać ich w podglądzie Markdown, a model je czyta.

What it does and when to use it

Use when you are investigating, fixing or changing code in a repository and your long-term memory is in Vornik: recall before reading code, design before changing it, keep memory current, and, where memory holds no design for the part being changed, write an as-built record of that part with the user, then store it. Never claims a design exists: it reads each recall's design_record count.

Original author description

Use when you are investigating, fixing or changing code in a repository and your long-term memory is in Vornik: recall before reading code, design before changing it, keep memory current, and, where memory holds no design for the part being changed, write an as-built record of that part with the user, then store it. Never claims a design exists: it reads each recall's design_record count.

How it works

  1. Investigate with memory first
  2. Design before changing code
Original SKILL.md headings · 76 lines, 716 words
SKILL.md preview
---
name: develop-with-memory
description: "Use when you are investigating, fixing or changing code in a repository and your long-term memory is in Vornik: recall before reading code, design before changing it, keep memory current, and, where memory holds no design for the part being changed, write an as-built record of that part with the user, then store it. Never claims a design exists: it reads each recall's design_record count."
---

In this harness the memory tools are `vornik_recall` and `vornik_remember`.
They search and store the whole project's memory, without a repository scope,
so `design_record` counts project-wide. There is no `delegate` or
`memory_correct` on a memory key: the user reviews an as-built record (rule
6), a reviewed record that is too large for `vornik_remember` is ingested by
the user with a Vornik client, and a wrong fact is reported to the user to
correct.

<!-- vornik-dev-guidance:start -->
# Developing with project memory

These rules are for changing code in a repository whose project memory is
kept in Vornik. They adapt to what that memory holds: many projects have few
written designs or none, and nothing here lets you claim a design exists.

## Investigate with memory first

1. **Recall before reading code** for a design, status, capability or "how
   does X work" question, an error or a surprising behaviour, and before your
   third read into one part of the code. Pass the repository's scope when
   your recall tool takes one. Read code to verify what a hit names, not to
   rediscover what memory holds: memory freezes facts when they are written.
2. **Read the recall's `design_record` before trusting a hit.** It counts the
   design-class hits (spec or decision) that carry this repository's scope
   among the hits returned (with no scope, the whole project's). Above zero,
   those hits are the design record for the question: check that a file,
   flag or function they name still exists. At zero, or when the field is
   absent, no design for this question was found in this repository's
   memory, whatever else ranked: hits from other repositories and notes are
   not its design. Say so, and work from the code. A design counts only when
   it was stored with this repository's scope. Every other hit (research,
   incident and companion notes) is a hint to verify, never a fact.
3. **Keep memory current.** When you change a design, re-ingest it in the
   same change, under the same path, so the old version stops being served.
   When a recalled fact contradicts the code, correct it (`memory_correct`
   with the hit's chunk id, then the corrected fact) instead of working
   around it.

<!-- vornik-design-first:start -->
## Design before changing code

Skip rules 4 to 8 when the user has said this project keeps its designs
elsewhere or does not work design first.

4. **Recall the design of the part you are about to change, before editing
   it.** Then one of three is true:
   - the design is silent on the change: add the change to it;
   - the design contradicts what the code does: correct it, dated, and mark
     the stale text where it stands;
   - there is no design (rule 2 found none): do not invent one, and do not
     skip this rule; offer the user rule 5.
5. **Write the as-built record, one part at a time.** With the user's
   agreement, write a design of what that part does today, from its code,
   its configuration and its observed behaviour, measured where you can (run
   times, sizes, counts), and say which statements you measured and which
   you inferred. Cover only the part the work touches, never the whole
   repository. Put it where the project keeps designs; when it keeps none,
   ask the user where. Head it "as-built, reverse-engineered on <date> at
   commit <sha>" and list the files it describes.
6. **Have it reviewed before anything relies on it.** Send the record and the
   files it describes to review (with `delegate`, the companion-doc-review
   workflow, which checks a document against the code it describes). Fix
   what is real and repeat until the review finds nothing blocking; after
   three rounds, ask the user. Without `delegate`, the user is the reviewer.
7. **Ingest only a reviewed record**, with this repository's scope and the
   file's path in the repository (the companion-rag-ingest workflow, or
   `remember` with class `spec` when it is small), so memory never serves an
   unreviewed draft as a design. Then make the change as a dated section of
   the same record, test first, and amend the record with what shipped.
8. **When you edit a file a record lists, amend that record in the same
   change** and re-ingest it. A record whose files changed after its commit
   is stale until someone checks it.
<!-- vornik-design-first:end -->
<!-- vornik-dev-guidance:end -->

Versions

RepositoryScoreLicence
This page · easeit-cz/vornik90AGPL-3.0
easeit-cz/vornik100AGPL-3.0
easeit-cz/vornik90AGPL-3.0

Database snapshot: 13.11.2026 · package version 8f44997af246 (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

  • 96vornik-broker
    Use whenever a request touches the user's mail, calendar, documents, accounts, money or any other private system. Vornik holds those credentials and does the work; you delegate and relay the result.
  • 100delegate
    Teaches Codex when to offload work to a vornik companion workflow instead of doing it inline. Use when the vornik MCP tools are available and the task is large, async-friendly, or belongs in project memory.
  • 100memory-benchmark
    Run, compare and debug the pi-cc-memory benchmarks (bench/ L1 store cost and L2 recall quality). Use when changing recall, the selector, indexing, locking or model-call code and you need before/after numbers; when asked "how well does recall work", "does it scale", "did this…
  • 100memory
    Deliberately search, add to, or curate your long-term memory with the `memory` tool — beyond the automatic recall/capture every turn already does. Use when asked who someone is or what they're working on, when you learn a durable fact worth keeping now, or when your notebook…
  • 100memory
    Deliberately search, add to, or curate your long-term memory with the `memory` tool — beyond the automatic recall/capture every turn already does. Use when asked who someone is or what they're working on, when you learn a durable fact worth keeping now, or when your notebook…
  • 100shodh-memory
    Persistent memory system for AI agents. Use this skill to remember context across conversations, recall relevant information, and build long-term knowledge. Activate when you need to store decisions, learnings, errors, or context that should persist beyond the current session.

Feedback and reports

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