[Chinese](README.md) | [English](README.en.md) # AK體 2.0 - Threads Algorithm Content Decision Skill `AK-Threads-Booster` remains the internal package name and install id. AK-Threads-Booster is an AI skill system for Threads creators who want better posting decisions, not just faster text generation. Its job is to help users: - find stronger topic angles faster - avoid obvious repetition and red-line mistakes - draft in a voice closer to their own - learn from real post performance over time It does not guarantee viral posts. It turns topic selection, drafting, analysis, prediction, and review into a repeatable system backed by the user's own Threads history. --- ## What Is New In 2.0 Version 2.0 turns AK-Threads-Booster from a single-post assistant into a fuller Threads content operating system. - **Low-token compiled memory**: converts the tracker into an account wiki, post feature index, topic clusters, next-move queue, exemplar bank, and voice fingerprint so daily runs can start from compact runtime memory. - **Next Move Engine**: recommends not just topics, but the account bottleneck the next post should address, such as stronger judgment, better discussion density, more saveable value, freshness recovery, or lower AI-tone risk. - **Voice + Cognitive + Draft Operating Pack**: `/voice` now builds a deterministic voice fingerprint first, then distills beliefs, tensions, anti-voice boundaries, and a `/draft` quick-reference pack. - **Local visual panel**: inspect account health, trends, top posts, topic distribution, compiled memory, and next-move signals with zero token use. - **Safe `/update` module**: checks for newer versions and can offer opt-in weekly update checks; it only fast-forwards clean repos and never overwrites local changes. - **Agent-generic packaging**: removes platform-specific install assumptions and exposes `AGENTS.md`, `SKILL.md`, and `agents/openai.yaml` for different agent environments. --- ## Who It Is For - creators already posting on Threads who want a steadier content process - users who want to know which topic types actually travel for their account - users tired of guessing what to post next - users who want their post history to become a usable decision asset If the goal is to choose better topics, improve distribution odds, and stop wasting posts on stale or repetitive angles, this skill is built for that. --- ## Core Modules ### `/topics` Find the next most worthwhile topic by combining historical performance, comment demand, self-repetition risk, and external freshness. ### `/draft` Generate a draft from the user's brand voice, style guide, and historical data. Before drafting: a freshness gate, a fact-check that **never overrides the user's own stated personal facts or event chronology**, and a research step that surfaces 2-3 angles the user may not have considered. After drafting: 3-5 targeted questions to sharpen the post. All dialogue is toggleable via `threads_booster_config.json` (`ask` / `always_on` / `always_off`). ### `/analyze` Run decision-first analysis on a finished draft. It checks algorithm red lines, upside drivers, suppression risks, style fit, and AI-tone traces. ### `/predict` Estimate likely 24-hour performance from comparable historical posts so expectations are anchored in data. ### `/review` Compare actual results against the prediction and write the learning back into the tracker. ### `/refresh` Update `threads_daily_tracker.json` through the Threads API when available, or through an authenticated browser automation environment when API access is not available. ### `/voice` Build a Brand Voice profile that is useful for drafting, not just descriptive. It combines engagement-weighted evidence, temporal shift, cognitive beliefs, tension pairs, anti-voice boundaries, and a `/draft` quick-reference pack. ### `/panel` Open a local zero-token dashboard for account status, recent trends, top posts, topic distribution, post search, compiled memory, and next-move signals. ### `/update` Check for newer versions and ask whether the user wants opt-in weekly update checks. Updates only run when the local repo is clean and can fast-forward safely. --- ## What Setup Produces After `/setup`, the working directory typically contains: - `threads_daily_tracker.json` - `style_guide.md` - `concept_library.md` - `brand_voice.md` after `/voice` - `compiled/account_wiki.md` - `compiled/account_state.md` - `compiled/next_move_queue.md` - `compiled/post_feature_index.jsonl` - `compiled/voice_fingerprint.md` / `.json` - `posts_by_date.md` - `posts_by_topic.md` - `comments.md` The tracker is the canonical file. Companion files and compiled memory are derived from it; they are useful runtime assets, not new sources of truth. --- ## Recommended Flow ### First-time setup ```text /setup /voice ``` This builds the tracker first, then deepens the brand-voice layer for better drafting. The newer `/voice` flow first builds a local voice fingerprint, then uses AI judgment for cognitive core, anti-voice boundaries, and the `/draft` quick-reference pack. `/voice` produces a **first-draft reference** `brand_voice.md`, not a verdict. An outside LLM always misses things the author knows about themselves. Expected usage: - Edit anywhere it feels wrong — your edits win. - Fill in the **Manual Refinements** section at the bottom with taboos, must-do rules, and "not me" phrases. - Re-runs of `/voice` preserve your edits and the Manual Refinements section. `/draft` treats Manual Refinements as hard constraints, ranked above other generated sections. It then reads Cognitive Core, the Quick-Reference Pack, Anti-Voice, and Voice Fingerprint. ### Before posting ```text /topics /draft /analyze ``` Use `/topics` to find the best next angle, `/draft` to create a starting point, and `/analyze` to pressure-test the finished draft before publishing. ### Skill Updates Use `/update` to check whether AK-Threads-Booster has a newer version. After a check, `/update` will proactively ask whether you want to enable weekly auto-update checks. If enabled, it only fast-forwards when the local repo is clean and safe to update. If local edits, local-only commits, or conflicts exist, it stops and reports the blocker instead of overwriting anything. ### After posting ```text /predict /review ``` This closes the loop and makes the next decision better. ### Inspect account status first ```text /panel ``` Or run locally: ```bash python scripts/panel_server.py --open ``` The panel is useful before spending tokens on deeper analysis. --- ## Data Sources Users can build the system from: - Threads Developer API token - Meta official export zip - existing JSON / Markdown / CSV - an authenticated browser automation environment logged into Threads - legacy tracker migration API access is optional, but it makes refresh much easier. --- ## Product Positioning This is not a "guaranteed viral post" tool. The durable promise is: **help creators find better topics faster and improve the odds that their posts are worth sharing, saving, and discussing.** That is a stronger and more honest product promise than claiming guaranteed results. --- ## Installation Give this GitHub repo to your agent: ```text https://github.com/akseolabs-seo/AK-Threads-booster ``` Agents that support repo instructions or skill directories can read `AGENTS.md` or `SKILL.md`, then route into `/setup`, `/voice`, `/topics`, `/draft`, `/analyze`, and the other modules. Environments that support OpenAI/Codex-style discovery can also read `agents/openai.yaml` as UI metadata. You can also clone it manually: ```bash git clone https://github.com/akseolabs-seo/AK-Threads-booster.git ``` Place the repo in the skill or agent-instructions directory used by the target tool. --- ## Directory ```text AK-Threads-booster/ |- SKILL.md |- AGENTS.md |- agents/ | |- openai.yaml |- skills/ | |- setup/SKILL.md | |- refresh/SKILL.md | |- analyze/SKILL.md | |- draft/SKILL.md | |- predict/SKILL.md | |- review/SKILL.md | |- topics/SKILL.md | |- voice/SKILL.md | |- panel/SKILL.md | |- update/SKILL.md |- knowledge/ | |- _shared/ | |- psychology.md | |- algorithm.md | |- ai-detection.md | |- data-confidence.md | |- chrome-selectors.md |- scripts/ | |- fetch_threads.py | |- parse_export.py | |- build_compiled_memory.py | |- build_voice_distillation.py | |- check_skill_update.py | |- panel_server.py | |- update_snapshots.py | |- update_topic_freshness.py | |- render_companions.py |- panel/ |- templates/ |- examples/ ``` --- ## License MIT License. See [LICENSE](./LICENSE).