[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:skillforge":3},"\u003Ch1>SkillForge v5.2\u003C\u002Fh1>\n\u003Cp>\u003Cstrong>From Art to Engineering: A Manifesto for AI Skill Creation.\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Ftripleyak\u002Fskillforge\u002FHEAD\u002Fassets\u002Fimages\u002F01-title.png\" alt=\"SkillForge\" \u002F>\u003C\u002Fp>\n\u003Chr \u002F>\n\u003Ch2>The Problem\u003C\u002Fh2>\n\u003Cp>The central challenge in AI development isn't a lack of ideas, but the inconsistent process of turning them into robust, reliable skills. Current methods are often ad-hoc, brittle, and difficult to scale—resembling more of an art form than a predictable engineering discipline.\u003C\u002Fp>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Ftripleyak\u002Fskillforge\u002FHEAD\u002Fassets\u002Fimages\u002F02-quality-gap.png\" alt=\"The Quality Gap\" \u002F>\u003C\u002Fp>\n\u003Chr \u002F>\n\u003Ch2>The Solution\u003C\u002Fh2>\n\u003Cp>\u003Cstrong>Quality is built in, not bolted on.\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cp>SkillForge is a methodology where rigor is integrated into every step of the creation process, from initial conception to final validation. It's a fundamental shift from reactive testing to proactive engineering.\u003C\u002Fp>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Ftripleyak\u002Fskillforge\u002FHEAD\u002Fassets\u002Fimages\u002F03-quality-built-in.png\" alt=\"Quality Built In\" \u002F>\u003C\u002Fp>\n\u003Chr \u002F>\n\u003Ch2>What's New in v5.2\u003C\u002Fh2>\n\u003Cp>v5.2 adds the Context Skill Advisor: proactive, evidence-backed skill suggestions with user-controlled Proactivity Levels.\u003C\u002Fp>\n\u003Ch3>Context Skill Advisor\u003C\u002Fh3>\n\u003Cul>\n\u003Cli>Proactive by default with install-time levels: \u003Ccode>off\u003C\u002Fcode>, \u003Ccode>quiet\u003C\u002Fcode>, \u003Ccode>balanced\u003C\u002Fcode>, and \u003Ccode>active\u003C\u002Fcode>\u003C\u002Fli>\n\u003Cli>\u003Ccode>balanced\u003C\u002Fcode> is the default level\u003C\u002Fli>\n\u003Cli>Supports global config plus project-level overrides\u003C\u002Fli>\n\u003Cli>Uses Session Context, Project Context, and Personal Context by default\u003C\u002Fli>\n\u003Cli>Uses Targeted Content Access: search broadly, then read only narrow relevant excerpts\u003C\u002Fli>\n\u003Cli>Runs from Advisor Checkpoints and Scheduled Background Advising\u003C\u002Fli>\n\u003Cli>Writes suggestions to an Advisory Queue and learns from Advisor State feedback\u003C\u002Fli>\n\u003Cli>Requires confirmation before invoking suggested skills\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3>Advisor Commands\u003C\u002Fh3>\n\u003Cpre>\u003Ccode class=\"language-bash\"># Configure proactive advising\npython scripts\u002Finstall_skillforge.py\n\n# Run a checkpoint from the current agent session\npython scripts\u002Fcontext_advisor.py checkpoint --cwd \"$PWD\" --text \"<brief current context>\"\n\n# Run scheduled advising and queue suggestions\npython scripts\u002Fcontext_advisor.py run --cwd \"$PWD\"\n\n# Review queued suggestions\npython scripts\u002Fcontext_advisor.py list\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Chr \u002F>\n\u003Ch2>What's New in v5.1\u003C\u002Fh2>\n\u003Cp>v5.1 builds on the v5.0 context-efficient redesign and adds stronger frontmatter support, hooks guidance, validation coverage, and packaging safety.\u003C\u002Fp>\n\u003Ch3>Context-Efficient Foundation (v5.0)\u003C\u002Fh3>\n\u003Cp>The foundation from v5.0 remains: \u003Cstrong>the context window is a public good.\u003C\u002Fstrong> Every line in SKILL.md competes with the user's actual work.\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>SKILL.md slimmed from 872 to 313 lines\u003C\u002Fstrong> (64% reduction)\u003C\u002Fli>\n\u003Cli>Deep dives moved to \u003Ccode>references\u002F\u003C\u002Fcode> where they're loaded only when needed\u003C\u002Fli>\n\u003Cli>Triggers moved into \u003Ccode>description\u003C\u002Fcode> field for pre-load routing\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3>Simplified Frontmatter\u003C\u002Fh3>\n\u003Cp>Skills now use only \u003Ccode>name\u003C\u002Fcode> and \u003Ccode>description\u003C\u002Fcode> in frontmatter. The \u003Ccode>description\u003C\u002Fcode> field is the primary triggering mechanism — it determines when a skill activates, so all \"when to use\" information belongs there.\u003C\u002Fp>\n\u003Cpre>\u003Ccode class=\"language-yaml\">---\nname: my-skill\ndescription: What this skill does and when to use it. Include trigger scenarios.\n---\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Ch3>Degrees of Freedom\u003C\u002Fh3>\n\u003Cp>A new design concept for matching instruction specificity to task fragility:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>High freedom\u003C\u002Fstrong> (text guidance) — when multiple approaches are valid\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Medium freedom\u003C\u002Fstrong> (pseudocode\u002Fparameterized scripts) — when a preferred pattern exists\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Low freedom\u003C\u002Fstrong> (exact scripts) — when operations are fragile and error-prone\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3>Scaffold Script\u003C\u002Fh3>\n\u003Cp>New \u003Ccode>init_skill.py\u003C\u002Fcode> creates rich skill templates with TODO placeholders, organizational pattern suggestions, and example resource files:\u003C\u002Fp>\n\u003Cpre>\u003Ccode class=\"language-bash\">python scripts\u002Finit_skill.py my-new-skill --path ~\u002F.codex\u002Fskills\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Ch3>Iteration as a Formal Step\u003C\u002Fh3>\n\u003Cp>Iteration is now built into Phase 3. Skills improve through real usage, not just synthesis panel review.\u003C\u002Fp>\n\u003Ch3>Extended Frontmatter + Hooks (v5.1)\u003C\u002Fh3>\n\u003Cp>v5.1 expands skill metadata support and documentation:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Extended frontmatter coverage for \u003Ccode>model\u003C\u002Fcode>, \u003Ccode>context\u003C\u002Fcode>, \u003Ccode>agent\u003C\u002Fcode>, \u003Ccode>hooks\u003C\u002Fcode>, and \u003Ccode>user-invocable\u003C\u002Fcode>\u003C\u002Fli>\n\u003Cli>New hooks integration guidance for \u003Ccode>PreToolUse\u003C\u002Fcode>, \u003Ccode>PostToolUse\u003C\u002Fcode>, and \u003Ccode>Stop\u003C\u002Fcode>\u003C\u002Fli>\n\u003Cli>Template updates for modern skill authoring defaults\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3>Validation + Packaging Hardening (v5.1)\u003C\u002Fh3>\n\u003Cp>v5.1 adds stronger guardrails for safe distribu\u003C\u002Fp>\n",1784240408082]