Most answers to « give my AI agent long-term memory » hand you an opaque vector store: embeddings in a database you can’t read, can’t diff, and can’t easily redact. I wanted the opposite — memory I can open in an editor, review in a pull request, and git blame line by line. That itch is what pushed OKF4net from a single library into a small toolkit over the last few months. It just reached v0.5, and this post is about what’s new — starting with the part I’m most excited about.
If you haven’t seen it before: OKF4net (docs & project site) is an independent, zero-dependency .NET implementation of Google’s Open Knowledge Format (OKF). OKF represents knowledge as a directory of markdown files with YAML frontmatter — cross-linked like a wiki, versioned in git. No database, no proprietary format: if you can cat a file you can read it, if you can git clone a repo you can ship it.
The headline: agent memory as plain markdown
OKF4net.Agents turns a bundle into tools and context for the Microsoft Agent Framework. OkfBundleTools exposes read, browse, graph, search, write, append-log, validate and more as function tools an agent can call directly. Layer OkfContextProvider onto the same instance and the agent automatically gets relevant bundle context injected into each turn — and, when you opt in, its exchanges captured back into the bundle as long-term memory:
using Microsoft.Agents.AI;
using Microsoft.Extensions.AI;
using OKF4net.Agents;
var tools = new OkfBundleTools("./my_bundle");
// Memory capture is opt-in (Disabled by default) — turn it on explicitly.
var provider = new OkfContextProvider(
tools,
new OkfContextProviderOptions { MemoryCapture = MemoryCaptureMode.Enabled });
AIAgent agent = chatClient.AsAIAgent(new ChatClientAgentOptions
{
ChatOptions = new ChatOptions { Tools = tools.GetTools() },
AIContextProviders = [provider],
});
var response = await agent.RunAsync("What do we know about orders?");
Here’s what makes it different. Capture is deterministic — no extra LLM call: after each turn, the last user message and the agent’s final response are appended to a single memory concept for the day (memory/<yyyy-MM-dd>), with a matching log.md change-history entry. Those writes go through the exact same validated, lock-protected, path-safe write path any other caller uses, and the captured text is blockquote-neutralized so a payload smuggled into a message can’t fake document structure.
The payoff is that memory is just files in your bundle. You can:
git blame a remembered fact to see exactly when and in which exchange it entered the memory,
diff memory across commits and redact a line with a normal edit,
point a second agent at the same directory with no export step,
and review the whole thing in a PR like any other change.
I’ll be honest about where v1 stands: this captured memory is bundle-global, unscoped, and opt-in — it carries no session/user/tenant key, so it’s meant for a shared, non-sensitive knowledge base, which is exactly why it ships disabled by default. If you need per-caller isolation, that lives one layer up in the catalog (below). I’d rather ship an honest, inspectable v1 than a magic box.
The rest of what’s new (v0.2 → v0.5)
The memory story is the headline, but the project grew in every direction since the first release. OKF4net now ships as five NuGet packages plus a couple of standalone tools:
A local MCP server, OKF4net.Mcp. The okf-mcpdotnet tool exposes a bundle to Claude Desktop or Claude Code over the Model Context Protocol — now with bundle auto-discovery, so it finds your knowledge base instead of making you wire a path. Recipes for other MCP-capable editors (VS Code, Cursor, Rider…) are on the issue tracker as good-first contributions.
A multi-bundle catalog, OKF4net.Catalog. A catalog.json manifest, a source resolver, and full-text search across many bundles — with read-only knowledge sources and writable memory tiers (session / user / tenant) for exactly the per-caller scoping the agent provider’s v1 memory doesn’t do.
Machine-readable CLI output. The Native AOT okf CLI (validate/info/index/graph/parse/fmt) gained a --json flag on validate and info, so a CI step can parse structured diagnostics instead of scraping text.
A producer that turns a repo into a bundle.producers/OkfProducer generates a validated OKF bundle straight from an existing repository (npm/NuGet/README detection so far). It’s an early walking skeleton, but it closes the loop: you don’t have to hand-author a bundle to try the ecosystem.
Through all of it, the core constraint held: OKF4net and its CLI have zero third-party runtime dependencies — base class library only, hand-written YAML-subset parser and link scanner included. That’s what lets the CLI ship as a single self-contained Native AOT binary with no runtime to install, and it keeps the barrier to contributing low: there’s no framework to learn before you can read the code. Much of OKF4net is built with AI assistance (Claude Code), with the OKF spec and an extensive test suite — including byte-exact golden CLI captures — as the ground truth every change has to satisfy.
Come contribute
OKF4net is open source under LGPL-3.0-or-later and deliberately welcoming to first contributions — no prior OKF knowledge needed. The good first issue label names the files to touch and the test that should go green, ROADMAP.md shows where it’s headed, and Discussions is the place to ask before you write any code. If you’d rather start by reading, the project site and docs are the friendlier way in. Your first PR is three commands away: dotnet build, dotnet test, dotnet format.
If « agents that remember things in files you can read » sounds useful, I’d love the help — and the feedback.
If you can cat a file, you can read the knowledge base. If you can git clone a repo, you can ship it. No vector database to stand up, no proprietary export format, no vendor lock-in — just a directory of markdown files with YAML frontmatter that a human can open in any editor and an agent can read with ReadFile. That’s the whole pitch, and it’s the reason I spent the last few weeks porting a Rust library to C# to get it onto .NET.
The format behind that pitch is Google’s Open Knowledge Format (OKF) v0.1, and the library is OKF4net (docs & project site) — a zero-dependency .NET
(C#, net10.0) implementation, plus an optional layer for wiring OKF bundles straight into agents built on the Microsoft Agent Framework. This post is the launch story: what OKF actually is, why I ported it instead of writing a wrapper, and what « zero dependency » really costs and buys you.
What OKF is
OKF defines a bundle: a directory tree of UTF-8 markdown files, where each file is a concept — a YAML frontmatter block followed by a markdown body. Concepts cross-link each other with ordinary markdown links, index.md files give you progressive-disclosure directory listings, and log.md files record date-grouped change history. The only hard conformance requirement is a non-empty type field on every concept; everything else — unknown types, unknown keys, broken links — has to be tolerated by a conformant consumer. It’s deliberately boring as a format, which is the point: the format is context here, not the pitch. The pitch is what you get to do with plain files — diff them, review them in a PR, grep them, back them up with nothing but git.
The port story
This repository used to ship a Rust implementation of OKF. I removed it at commit d20343c — but only after proving, file by file and command by command, that the C# port produced byte-identical output. tests/fixtures/golden/ holds five golden captures taken directly from the Rust binary’s stdout — validate, info, graph --dot, fmt, and index — against a shared example bundle, and the C# CLI is diffed byte-for-byte against every one of them in CI. As of today the full suite passes end-to-end, including five byte-exact golden CLI comparisons against the original captures.
I’ll say the quiet part out loud: this port was AI-assisted, done largely with Claude Code driving the migration file by file, spec section by spec section, with the golden fixtures as the ground truth it had to match exactly. I think that’s worth stating plainly rather than glossing over — a byte-exact port across languages is a fairly mechanical, well-specified translation task with an unambiguous pass/fail signal (does the output match the captured bytes, yes or no), which is exactly the kind of task where an AI pair-programmer earns its keep and where you can trust the result because you can verify it byte-for-byte rather than having to take anyone’s word for it. The interesting design decisions — the YAML subset, the permissive-loading philosophy, the two-tier validation split — came from following the spec and the Python reference implementation; the AI assistance was in the grinding, get-every-byte-right execution, not the architecture.
Show, don’t tell
Here’s the library, loading a bundle and running a conformance check:
using OKF4net;
var bundle = Bundle.Load("./my_bundle");
Console.WriteLine($"{bundle.Count} concepts");
// Conformance check (§9).
var report = BundleValidator.Validate(bundle);
if (report.IsConformant)
{
Console.WriteLine($"conformant with OKF v{OkfSpec.Version}");
}
// Traverse the cross-link graph.
var id = ConceptId.Parse("tables/orders");
foreach (var link in bundle.LinksFrom(id))
{
Console.WriteLine($"{id} -> {link.Target} (exists: {link.Exists})");
}
Bundle.Load never aborts on a malformed concept file — it collects parse failures into bundle.ParseErrors and keeps walking the tree, because a knowledge base that one bad file can take down entirely is a bad knowledge base.
And here’s the CLI, which is the same tool the Rust binary used to be, invocation-for-invocation:
okf validate exits non-zero on a non-conformant bundle, so it drops straight into a CI step. The CLI ships as a self-contained, Native AOT single-file binary — no .NET runtime install required on the machine that runs it.
The agents angle
The reason I care about this format enough to port a whole library for it is OKF4net.Agents, which turns an OKF bundle into tools and context for the Microsoft Agent Framework. OkfBundleTools wraps one bundle root and exposes nine function tools — read, browse, graph, search, write, append-log, regenerate-indexes, validate, changes-since — that an AIAgent can call directly:
var tools = new OkfBundleTools("./my_bundle");
AIAgent agent = chatClient.AsAIAgent(tools: tools.GetTools());
var response = await agent.RunAsync("Search the bundle for concepts about refunds.");
Layer OkfContextProvider onto the same tools instance and, opted in explicitly, an agent’s exchanges get captured as long-term memory — one markdown concept per UTC day, written through the same validated, lock-protected write path the tools use, plus a matching log.md entry. That’s the part I think is genuinely different from the usual answer to « give my agent memory »: instead of an opaque vector store you can’t audit, memory is a markdown file in a git-tracked directory. You can open it, diff it across commits, redact a line, or point a second agent at the exact same directory with no export step. It’s not a fit for every use case — the README is upfront that v1 memory is bundle-global and unscoped, so it’s opt-in and meant for a shared, non-sensitive bundle rather than a multi-tenant deployment — but for a single team’s shared knowledge base, « memory you can git blame » is a real capability, not a slogan.
Design choices
The whole library — OKF4net and OKF4net.Cli — has zero third-party runtime dependencies: no YAML library, no CLI-parsing package, nothing. It has its own documented YAML subset parser (frontmatter is scalars, lists, and shallow maps — no anchors, no tags, no multi-document files, and it says so with a clear error if you hand it those), its own markdown link scanner, and its own argument parsing, all on top of the .NET base class library. That constraint is what makes the CLI publishable as a single-file Native AOT binary with no runtime to install, and it’s what keeps the barrier to contributing low — there’s no framework to learn before you can read the code. OKF4net.Agents is the one exception, since talking to Microsoft.Agents.AI requires depending on it; everything else stays dependency-free by design, enforced project by project. The project also ships OKF4net.Catalog, a local multi-bundle catalog with search-by-source resolution, and OKF4net.Mcp, an MCP server that plugs a bundle straight into Claude Desktop or Claude Code — so agents and tools have a ready path to discover and query bundles without writing that plumbing themselves.
Come contribute
OKF4net is young and I’d rather it stay welcoming than gate-kept. You don’t need any prior OKF knowledge to help — the good first issue label names the files to touch and the test that should go green when you’re done, ROADMAP.md lays out where the project is headed, and Discussions is the place to ask a question before you write any code. The project is licensed LGPL-3.0-or-later, and the bar to your first PR is exactly three commands: dotnet build, dotnet test, dotnet format. If any part of « knowledge bundles you can cat and agents that remember things in files you can read » sounds useful to you, I’d love the help — and the feedback.
Parts 1 and 2 covered the model and the full import/configuration procedure. This last part is about timing and running it in production: the day-before / day-of / stabilization timeline, scheduled archiving, backups, GDPR re-activation, and rollback.
The single most under-estimated fact of this whole migration:
The migration starts the day before, not on go-live day. The 05:00 go-live is only the tracking switch. All the heavy lifting — import, configuration, first archive — happens the evening before.
When
What
Duration
Day-1, afternoon
Import + configuration + Tag Manager + rehearsal
≈ 4 h (incl. ≈1h45 import)
Day-1, evening
First full archive
several hours on a large history
Day, before 05:00
Tracking cutover
≈ 30 min
Phase A — Day before (Day-1)
This phase runs everything from Part 2, in order, ending with the first full archive. The steps that specifically belong to the day-before rehearsal:
A1–A2. Sanity gates. Containers are up (podman ps), and the image version is ≥ the dump’s version_core. If the image is older, stop here — core:update will refuse and nothing downstream matters.
A3–A8. Import and wire up. Detect the real dump format, derive the prefix, create the DB/user, verify the DB is empty, import (in tmux), verify the import and restore durability, write config.ini.php. (All detailed in Part 2.)
A9. core:update → expect « Everything is already up to date ».
A10. Tag Manager → activate, regenerate containers, confirm a 200.
A11. Inventory then disable scheduled reports — save the list first, because you’ll need it to re-enable exactly the same ones later:
$DB_EXEC="podman exec -i mariadb"
$DB_EXEC mariadb -u root -p"<ROOT_PWD>" matomo -e \
"SELECT idreport, idsite, login, description, period FROM report WHERE deleted = 0;" \
| tee /var/backups/matomo/active-reports.txt
$DB_EXEC mariadb -u root -p"<ROOT_PWD>" matomo -e \
"UPDATE report SET deleted = 1 WHERE deleted = 0;"
deleted is a reversible flag, not a real delete. Keep active-reports.txt — it’s the only record of which reports to bring back.
A12. Disable the GDPR purge during migration — to rule out any deletion concurrent with cutover:
$DB_EXEC mariadb -u root -p"<ROOT_PWD>" matomo -e \
"UPDATE \`option\` SET option_value='0' WHERE option_name='delete_logs_enable';"
$APP_EXEC ./console core:clear-caches
A13–A14. Configure geolocation (UI) and generate security files.
A15. Persistence test — do not skip this. This is the check that catches the silent volume trap before it costs you. Restart the app container and confirm everything survived:
If the config disappears or the container falls back to 404, the volumes are not persistent → stop and fix with your infra team before any cutover. This is the trap that breaks everything silently on the first container recreation.
A16. First full archive — the evening of Day-1. Run it manually, in tmux:
tmux new -s archive
time $APP_EXEC ./console core:archive --url=https://analytics.example.com
Expect it to end with Done archiving!. An exit code 1 alongside Done archiving! is normal here — it comes from a failed report send (no SMTP yet), not from an archiving failure. This must run the evening before; started on go-live morning it won’t finish in time and reports would be slow and incomplete when users log in.
A17. Functional rehearsal — walk the go/no-go checklist below.
The go / no-go checklist
Before cutover, confirm:
[ ] Image on the pinned build; core:version ≥ dump’s version_core.
[ ] core:update ran without error.
[ ] Superuser login works (Cloud credentials); 2FA works.
[ ] Historical data visible (a past period renders).
[ ] Site main_url values updated (no leftover Cloud URLs).
[ ] Geolocation active for new traffic.
[ ] Scheduled reports disabled, and the idreport list saved.
[ ] Single collation — the collation query returns exactly one row.
[ ] 🚨 Tag Manager active and containers served (curl → 200, not 404).
[ ] config.ini.php, plugins, GeoIP and js/container_*.js on persistent volumes — verified by restarting the container.
[ ] A test hit shows up in real time.
[ ] Archive timer active + first run OK.
[ ] Monitoring in place (timer failure, disk space).
Some items are deliberately not satisfied at cutover and that’s fine — track them, don’t tick them: SMTP not configured, scheduled reports disabled, premium plugins absent, GDPR purge disabled, restorable backup tested. They’re decisions, not failures. Never tick them « to look clean » — a successful core:test-email during rehearsal would mean SMTP is live, which means reports can go out, exactly what you’re avoiding.
Phase B — Go-live day, before 05:00 (~30 min window)
B1. Confirm the day-before archive finished.
B2. Re-check the Tag Manager container serves a 200 (rerun A10 if it’s 404).
B3. Switch the tracking — the irreversible move on the sites. First, separate your two populations, because they switch differently:
# Sites WITH a Tag Manager container -> switch the container URL
$DB_EXEC mariadb -u root -p"<ROOT_PWD>" matomo -e \
"SELECT s.idsite, s.name, c.idcontainer FROM \`site\` s
JOIN tagmanager_container c ON c.idsite = s.idsite AND c.status='active'
ORDER BY s.idsite;"
# Sites WITHOUT a container -> switch the classic tracking code
$DB_EXEC mariadb -u root -p"<ROOT_PWD>" matomo -e \
"SELECT idsite, name FROM \`site\`
WHERE idsite NOT IN (SELECT idsite FROM tagmanager_container WHERE status='active');"
Then, on the sites:
Update the classic tracking code (matomo.js / matomo.php URL) to analytics.example.com — the sites without a container first, and anywhere the snippet is hard-coded.
Update the Tag Manager container URL on the sites that use one.
Or switch DNS if you keep the same hostname — then no URL changes are needed.
« All sites are reporting » is not a sufficient check — it doesn’t prove the Tag-Manager-published sites are covered. Check the two populations separately in B4.
B4. Verify real-time collection, per population:
$DB_EXEC mariadb -u root -p"<ROOT_PWD>" matomo -e \
"SELECT s.idsite, s.name,
CASE WHEN c.idsite IS NULL THEN 'direct code' ELSE 'Tag Manager' END AS mode,
COUNT(v.idvisit) AS visits_30min
FROM \`site\` s
LEFT JOIN (SELECT DISTINCT idsite FROM tagmanager_container WHERE status='active') c
ON c.idsite = s.idsite
LEFT JOIN log_visit v ON v.idsite = s.idsite
AND v.visit_last_action_time > NOW() - INTERVAL 30 MINUTE
GROUP BY s.idsite, s.name, mode ORDER BY visits_30min ASC;"
If all Tag Manager sites are at 0 while direct-code sites report, the container isn’t served or the URL wasn’t switched → revisit B2/B3. A single site at 0 isn’t necessarily a failure — low-traffic sites at 5 AM legitimately show zero; compare to each site’s usual volume, not to zero.
B5. Verify real IPs — the reverse-proxy trap. Don’t rely on counting distinct IPs. Test against a known IP:
# 1) From the test machine, note its public IP:
curl -s https://ifconfig.me ; echo
# 2) Generate a visit from that machine on a tracked site.# 3) Confirm Matomo recorded THAT IP, not the proxy's:
$DB_EXEC mariadb -u root -p"<ROOT_PWD>" matomo -e \
"SELECT INET6_NTOA(location_ip) AS ip, COUNT(*) AS visits FROM log_visit \
WHERE visit_last_action_time > NOW() - INTERVAL 15 MINUTE \
GROUP BY location_ip ORDER BY visits DESC;"
The IP from step 1 must appear. If every visit carries the proxy IP (or an internal 10.x / 172.16-31.x / 192.168.x), the forwarded-for headers aren’t being applied → fix proxy_client_headers and core:clear-caches. Fix immediately — visits collected meanwhile are falsified and unrecoverable. The header name must match what your proxy actually sends (X-Forwarded-For usually, sometimes X-Real-IP).
Phase C — Right after cutover (H+0 to H+2)
C1. Watch real-time for ~1h; confirm all sites report.
C2. Enable scheduled archiving — on a container host, prefer a systemd timer over cron; it shares the container’s mode (rootful/rootless) and logging.
core:archive does more than archive. At the end of each run it triggers the scheduled tasks: emailing reports and the GDPR log purge. Without this timer, neither the reports nor the purge ever run. To trigger them in isolation: ./console scheduled-tasks:run.
C3. Inform users: new URL, unchanged credentials (passwords and 2FA migrated), email reports temporarily suspended, and any premium features currently unavailable.
Phase D — Stabilization (Day+1 to Day+7)
Order is imposed — do not invert it. Reports have been disabled since A11. Wire SMTP first (D1) so you can test it empty and safe, then re-enable reports (D2) knowingly. The reverse order — reports active before a working SMTP — blasts emails on the next archive run.
D1. SMTP — configure the relay, then validate empty:
$APP_EXEC ./console core:test-email
Expect the test email to arrive. Reports are still disabled, so no mass send is possible yet. If it fails, fix the relay before D2.
D2. Re-enable scheduled reports — only the ones saved in active-reports.txt, and only after D1 passes:
$DB_EXEC mariadb -u root -p"<ROOT_PWD>" matomo -e \
"UPDATE report SET deleted = 0 WHERE idreport IN ( /* ids from active-reports.txt */ );"
Never run a global UPDATE report SET deleted = 0 — it would resurrect reports that were intentionally deleted while on Cloud.
D3. Re-enable the GDPR purge:
$DB_EXEC mariadb -u root -p"<ROOT_PWD>" matomo -e \
"UPDATE \`option\` SET option_value='1' WHERE option_name='delete_logs_enable';"
$APP_EXEC ./console core:clear-caches
Leaving it off makes the database grow indefinitely and steps outside your declared retention — it’s a compliance control, not an optimization.
D4. Backups — set up database + application volume, and test a restore. The application volume matters: it holds config.ini.php, plugins and the GeoIP database — restoring only the database won’t bring the service back. A reference logical dump:
Rotate on a separate target. A backup that has never been restored is not a backup.
D5. Confirm durability is restored (SELECT @@innodb_flush_log_at_trx_commit; → 1).
D6. Cancel the Cloud subscription — point of no return. Only after: a conclusive observation period, a backup successfully restored at least once, and any needed export of the data gap (below).
The data gap — a decision to make explicitly
The dump is a snapshot at time T. Between the dump and the cutover, Cloud keeps collecting. A common, defensible decision is to accept the gap: don’t replay a final dump at go-live; the visits between the dump’s last data and the tracking switch stay only in Cloud and aren’t recovered on-premise.
Consequences to keep in mind:
✅ The cutover window stays short — no ~2h import to fit in, just switching tracking code / DNS.
✅ The pinned image stays valid (it matches the already-imported dump) — no version re-qualification.
⚠️ The gap grows over time — its size is the interval between the dump’s last data and the cutover date. The later the go-live, the longer the missing period. That’s the one argument for cutting over sooner rather than later.
💡 Cloud remains readable until cancellation — if the missing period ever needs analysis, export or consult it from Cloud before cancelling (D6 is the point of no return).
Rollback
As long as Cloud is not cancelled, rolling back is quick:
Restore the old tracking code and Cloud container URLs on the sites.
Confirm collection resumes on Cloud.
The on-premise instance can stay in place for analysis.
No on-premise data is lost — the imported database stays intact. Keep both the original dump and an on-premise backup before any destructive operation.
Part 2 — the full procedure: prepare MariaDB, inspect and import, wire up config.ini.php, core:update, and post-migration configuration.
Part 3 — timing and operations: the day-before / day-of / stabilization runbook, archiving, backups, GDPR, and rollback.
The recurring lesson across all three: the import is the easy part. What breaks a Matomo Cloud→self-hosted migration is everything the dump doesn’t carry — and the fact that most of it fails silently, well after the migration looks done.
