BLOG
Writing
Long-form essays on engineering decisions — how teams structure them, score evidence, and stop forgetting why they were made.
-
NEW How to start using FPF: the minimal protocol
Three steps to try Anatoly Levenchuk's pattern language on your own project. With two Name Card examples and a list of traps that are easy to fall into when working with an LLM.
2026-05-26
-
NEW FPF: a map for thinking when you're not working alone
Anatoly Levenchuk's pattern language for teams with humans and AI agents - in plain English with examples for PMs, engineers, and everyday people.
2026-05-26
-
NEW Where FPF helps: 25 situations across 12 fields
From a payments API to a family decision about moving countries - concrete situations where Anatoly Levenchuk's FPF patterns turn confusion into working structure. Each case maps to a specific FPF pattern.
2026-05-26
-
NEW Process over model: three theses on why AI without guardrails breaks code
Three aphorisms I repeat more than the rest: 'PRD = WHAT, not HOW', 'CLAUDE.md is one agent's constitution, BMAD is the team's', 'without a process, AI raises code complexity'. The last one isn't rhetoric - MSR 2026 gives numbers: 42.7% of agent commits raise cyclomatic complexity, 56.1% lower the Maintainability Index, +39% cognitive complexity over time.
2026-05-25
-
NEW The full cycle on a single feature - from a one-liner in chat to a decision that opens itself for review a year later
Six months ago, one line in chat: 'we need streaming for model responses.' Today, a connected set of six files marked active, with measured trust and explicit review conditions. The series capstone: the path from that line to a decision, in seven steps on one real task.
2026-05-24
-
NEW POS terminal for methodologies - what Forgeplan is and why hold four ways of working in one plane
A waiter doesn't keep the price list, dish ingredients, taxes, and loyalty points in their head. They press buttons on a terminal, and the terminal remembers everything. Forgeplan is a POS terminal for methodologies: four ways of working in one plane, so you pick the content and the mechanics stay out of the way.
2026-05-23
-
NEW Spec is not PRD, it's the taste test - intent vs behavior, and why the difference is huge in practice
A recipe can sound delicious and still be unworkable. The final taste of a dish is only verified by eating it. A PRD is the recipe. A Specification is the taste test: verifiable system behavior, a delta format for changes like commit messages, a dependency graph between related specs.
2026-05-22
-
NEW An architect doesn't start with blueprints - four phases any non-trivial feature passes through, consciously or not
Between a one-line Slack message and a working feature there are four phases: expanding the solution space, breaking it into parts and edge cases, choosing an option, writing the contract document. The only question is whether a team runs these phases in a document in one day - or in code over two weeks of untangling consequences.
2026-05-21
-
NEW The motivation section for an architectural decision - making a record that opens itself for review a year later
A judge writes which evidence was considered, which was rejected, and why - so the case can be reopened a year later. Architectural decisions need exactly the same discipline. I break down the six sections of a DDR, with special attention to the one that's skipped most often: conditions for reopening.
2026-05-19
-
NEW Averages lie: why the trust in a decision is capped by its weakest piece of evidence, not the average of all pieces
A vendor whitepaper and a Slack screenshot from a trusted colleague can score the same 'average quality' - but betting a decision on one vs the other is a different risk entirely. I break down R_eff: three axes of evidence, the weakest-link rule, context matching - and why one benchmark you ran yourself beats ten blog posts from strangers.
2026-05-18
-
NEW Before you fix it, write three versions of the cause: the detective's move that turns 30 minutes of guesswork into 5 minutes of accurate diagnosis
A good detective doesn't lock onto the first version of events. They propose three or four, describe what evidence each predicts, then go check. The same move - ADI - turns the hasty wrong fix into ten minutes of honest diagnosis. I walk through two real cases: a conversion drop and a slow search.
2026-05-16
-
NEW The decision graveyard: why six months later nobody remembers why you picked one auth flow over another
Teams accumulate invisible debt not in their code, but in forgotten architectural decisions. Notion and Confluence don't fix this - you need a different format. I break down what this debt looks like on real cases and what a decision format must contain to still be useful a year later.
2026-05-14
-
R_eff: the math behind "trust = weakest link" in evidence-based decisions
ForgePlan scores every decision with R_eff = min(evidence_scores) - never the average. This post unpacks why the minimum, how individual evidence scores are computed, and how to read a real R_eff number in forgeplan score.
2026-05-12
-
Markdown as source of truth, LanceDB as derived index - and the 32 violations that almost killed the invariant
How ForgePlan's load-bearing ADR-003 invariant got compile-time-enforced: 4 audit rounds, 56 findings, a pub(crate) lockdown across 32 call sites, and why the boring discipline matters more than clever architecture.
2026-05-10
-
Harness engineering for AI coding agents - why your prompt isn't the problem
If your AI agent ships a feature with the wrong logic and a green test suite, the failure is in the harness - not the prompt. A walkthrough of the five harness subsystems using ForgePlan as a concrete implementation, with code examples and honest gap list.
2026-05-08
-
I built a "git for decisions" and use it to teach engineers - six months in
Six months of dogfooding ForgePlan - a local-first CLI that treats engineering decisions like first-class artifacts. Honest report: the methodology, a war story about a tool that lied, and what AI-native actually means in practice.
2026-05-05
-
Forgeplan: agent harness for AI coding agents (Rust + LanceDB)
Forgeplan is a local-first Rust CLI + MCP server that constrains where an AI coding agent can write, what state it must read, what evidence it must produce, and what gates it must pass. Markdown source of truth, compile-enforced invariant, weakest-link trust scoring, hint protocol contract.
2026-05-03
-
ForgePlan: an agent harness disguised as a decision framework
Six months of building a decision framework - then realising I'd built an agent harness. How ForgePlan maps to the five harness subsystems, the PROB-034 bug that taught me a harness must catch itself lying, and three honest gaps still on the roadmap.
2026-04-28
-
The decision graveyard: why my AI restarts the same argument every session
When AI breaks something, the problem almost never is the model. The problem is that the model has nowhere to put the decision. Six months of practice with ForgePlan: four ideas, three gaps, one star on GitHub.
2026-04-22
-
Building an agent harness in the open: six months of lessons
A harness is everything around the model except its weights. Six months of open ForgePlan development: five harness subsystems, how ADR-003 almost killed markdown as source of truth, and three roadmap holes I'm calling out publicly.
2026-04-15
-
ForgePlan: where your team's decisions actually live, if they're not lost
Six months of open development on a tool that turns team decisions into first-class git artefacts. What a decision graveyard is, how ForgePlan works, and why the repo has one star on GitHub - an honest report.
2026-04-05
-
Welcome to the Forgeplan blog
This blog will document decisions, releases, and methodology insights from Forgeplan development.
2026-03-25